The core argument: Mathematical notation is a language with its own grammar and spoken register. Secondary algebra students — especially multilingual learners — cannot acquire that register through exposure alone. And mathematics itself is not a cultureless, authorless system: it is the accumulated product of human civilizations across thousands of years. MASL makes both of these claims instructional. It provides structured activities for acquiring the spoken register of symbol sense, and it grounds each unit in the historical and cultural context that produced the mathematics students are learning.

Part 1 — ML Supports

Nine structured supports for multilingual learners, built into every lesson at the point of instruction — not as a separate pullout.

Core 1. MASL Card Sorts — Notation Language Scaffold
Every mathematical expression gets a three-card set: Math (symbol), We Say (spoken phrase), and Meaning (English sentence). Students physically match cards, connecting the notation they see to the language they use and the meaning they need. Alternate pronunciations are shown explicitly (e.g., "b i" or "b times i") so no student has to guess what to say aloud.
Core 2. Language Writing Standards
All student-facing text follows strict ML-first conventions: present tense, active voice, imperative mood, 12 words or fewer per instruction, one term per concept (no synonym switching), no idioms, no minimizing language, no contractions. Every math term is defined at first use.
Core 3. Sentence Frames
Any slide that asks students to write or speak a response includes a fill-in-the-blank frame labeled Try Saying. Students always have a linguistic model for the expected output — they never face a blank page. Frames never presuppose the answer.
Core 4. KLU Directive Verb + Frame Color Coding
The directive verb in a problem prompt (Explain, Describe, Argue, Recount) and its matching sentence frame share a color. The color is the retrieval cue: if a student can find the verb, they can find the scaffold.
Core 5. How To Reference Slides
After every teacher-led synthesis of a procedural skill, a click-to-reveal step-by-step reference table (Step | What to Do | Example) stays on screen during independent work. MLs can reference the procedure without holding it in working memory.
Core 6. Amplify CL Feedback
Wrong-answer feedback never reveals the answer — it identifies where to look and restates the setup as a question. This prompts re-examination in plain language rather than leaving a student stranded at "incorrect."
Core 7. Accessibility Floor
All student-facing text is 1rem (16px) minimum. Color and shape are always paired — never color alone. Contrast meets WCAG 2.1 AA throughout.
Math as Culture 8. Historical Readings — Engage Phase
One short historical/cultural reading per unit (150–300 words, Lexile 800–950) is placed at the start of the unit as the Engage moment. Each reading names a specific civilization, time period, and mathematician, and makes the connection to the mathematics students are about to study explicit. Readings are written at ML-accessible reading levels with pronunciation guides for unfamiliar proper nouns. They do not reduce non-Western contributions to "they gave us" — they convey complete mathematical traditions.
Math as Culture 9. Reflective Writing — Read → Reflect Beat
Immediately following the historical reading, a single structured reflection prompt asks students to connect, react, extend, or identify. The prompt is not a comprehension check. It asks students to do genuine intellectual work: what does knowing this history change about how you see this notation? Every prompt includes a Try Saying sentence frame. The Read → Reflect beat takes 10 minutes total (5 min read, 5 min write) and is its own instructional moment — never combined with activity time.
The underlying logic: MLs need the same math content with more explicit language support at every layer — notation, pronunciation, procedure, and writing. MASL provides that support at the point of instruction. And all students — MLs and native speakers alike — are better served when mathematics is presented as a human and cultural achievement, not an authorless set of rules.

Part 2 — Full Feature Set

All features of the MASL program — developed, in active use, and in progress.

A Math as Culture Layer New

Historical Reading
  • One per unit — placed on the first slide, before any math content
  • 150–300 words; Lexile 800–950; maximum 25 words per sentence
  • Required structure: (1) a human problem, (2) a cultural context with named civilization and person, (3) the mathematical idea that resulted
  • Pronunciation guides for unfamiliar proper nouns (e.g., "Al-Khwarizmi (al-KWAH-riz-mee)")
  • No idioms; no figurative language unless immediately explained; past tense for events, present tense for persistent mathematical ideas
  • Attribution block at bottom of slide; original passages attributed "Written for MASL · Math as Culture, Unit N"
  • Phase badge: Math as Culture · Reading (teal)
Reflection Prompt
  • One prompt only — three parts: anchor sentence, question, sentence frame
  • Scoped to a 3–5 sentence written response
  • Prompt types: connect, react, extend, or identify — never comprehend or summarize
  • No hero framing, no guilt framing, no forced cultural connection, no presupposed emotion
  • Sentence frame labeled Try Saying; frame starts in student voice ("I think…", "What surprised me was…"); frame never presupposes the answer
  • Phase badge: Math as Culture · Reflect (teal)
Placement rule: The Read → Reflect beat occupies two consecutive slides. Never place the reading mid-activity. Never combine reading and prompt on one slide. The beat needs clear entry and exit — it is its own instructional moment, not a warm-up to skip.
Cultural Representation Standard
Name the civilization and person specifically. "Indian mathematicians" is too vague. "Brahmagupta, a 7th-century mathematician working in what is now Rajasthan" is specific.
Do not reduce non-European contributions to "they gave us." These were complete, sophisticated mathematical traditions.
Do not use "primitive" or "ancient" as a synonym for "less advanced." Describe historical context without ranking it.
Do not center Europe as the default. When European mathematicians resisted an idea (negative numbers, imaginary numbers), name the resistance and explain it — do not treat European skepticism as the standard.
Accuracy over drama. Do not invent dialogue, emotions, or motivations not in the historical record.

B Content Pipeline

AccessIM Scraper
A Playwright-based automated scraper that logs into AccessIM (Illustrative Mathematics) with teacher credentials, navigates the lesson structure, and extracts full lesson content — activity narratives, vocabulary, problems, math expressions, and assets — into a canonical JSON schema. Modules: auth, navigate, extract, parse, output, assets, math-decoder, math-utils. Two parallel math rendering systems handled: MathJax (LaTeX from DOM annotations) and STIX SVG (server-rendered paths decoded glyph-by-glyph via math-decoder.js).
Lesson JSON Schema
Canonical data format consumed by all downstream generators. Fields include objectives, vocabulary (with masl_phrase), activities, problems, math_elements (latex + unicode + svg), assets, masl_instances (symbol / spoken / sentence triples, flagged for review), and scrape metadata.

C Slide System

Lesson Slides
HTML5 slide decks — self-contained single-file HTML, no build step, no framework, no server required. Fixed nav bar (previous / next / teacher toggle), keyboard navigation, slide counter.
  • Slide types: Launch, Work Time, Activity Synthesis, MASL Language & Notation, Math as Culture · Reading, Math as Culture · Reflect, Lesson Synthesis, Exit Ticket, Cool-Down, Homework, How To Reference
  • Two layout variants per lesson: single-day IM format; two-day adaptive format with Day 1/2 separator and nav indicator
Teacher Mode
Toggled by clicking the teacher button or pressing T. Reveals teacher notes and Amplify accordion panels. Hidden from students when projected. Teacher notes exempt from the 1rem font floor — read up close, not from the back of the room.
Phase Badge System
Every slide has a color-coded phase badge (top-left, always outside vertical centering). Warm-Up: amber; Launch: blue; Work Time: indigo/purple; Synthesis: orange; MASL: teal; Math as Culture: teal; Lesson Synthesis: amber; Exit Ticket/Cool-Down: violet; Reference/How To: indigo.
How To Reference Slides — Click-to-Reveal
Inserted after every Activity Synthesis for a procedural skill. Three-column table (Step | What to Do | Example) starts fully hidden. Teacher clicks each row during modeling; all rows remain visible afterward for student reference. Stays on screen through subsequent work time, exit ticket, or cool-down.

D Card Sort System

Print / Static Card Sort
Printable HTML with cut lines, color and shape coding (★ Math / ▲ We Say / ● Meaning), and a fold-and-separate answer key. Given cards marked "(given)".
Interactive Drag-and-Drop Card Sort
Pure HTML5 + vanilla JS. Card pool, drop zones, touch support, bump behavior (same-type card in occupied zone returns existing card to pool with orange flash animation), responsive zone sizing.
Partner Card Sort Variant
Two-player variant where each partner holds a different card subset, requiring communication to complete the sort. In progress.

E Amplify Activity Builder Integration

CL Script Generation
Every slide with a typed student response includes a ready-to-paste Amplify Computation Layer script in the teacher note. Scripts cover all four spacing variants per answer value, anticipated wrong-answer branches with targeted feedback, and a generic fallback. Teacher notes never contain answer keys — correct answers live inside CL feedback strings only.
Feedback Tone Standard
Wrong-answer feedback prompts re-examination rather than revealing the answer. An opener bank (twelve openers: "Almost!", "You're close —", "I see what happened —", etc.) is assigned intentionally per branch so repeated submissions hear different language. Generic fallback describes the approach and ends with a gentle open question.

F Color and Accessibility System

Semantic Color System
Colors encode roles, not content: primary form (blue), secondary/equivalent form (green), equivalence callout (amber), student fill-in (purple). Teal reserved exclusively for the spoken/linguistic layer and Math as Culture. Never reassign a color mid-activity.
WCAG 2.1 AA Compliance
  • Minimum 1rem (16px) for all student-facing text including phase badges, timing chips, and card labels
  • Contrast: ≥4.5:1 for normal text; ≥3:1 for non-text elements (borders, arrows, SVG strokes)
  • Color and shape always paired — never color alone
  • Radical notation standard: always parentheses inside radical symbols (√(a), not √a) to prevent ambiguity

G Authoring Standards and Documentation

Activity Writing Standards (activity_writing_standards.md)
Source-of-truth for all student-facing language. 10 sections covering voice, sentence structure, vocabulary, card labels, directions, math notation in text, print materials, accessibility, Amplify CL, and How To slides.
Math as Culture Writing Standards (math_as_culture_writing_standards.md) New
Standards for authoring historical readings and reflection prompts: length, narrative structure, voice, accessibility, cultural representation, attribution, prompt types, sentence frame rules, slide layout, timing, and quality checklist.

H In Progress / Planned

Planned Pacing Timers + Student Digital Clock
Slide-level and section-level pacing timers, color-coded by proximity to target time. Student-facing digital clock shown during timed work segments, teacher-adjustable in real time. Ships with the slide generator — no backend required.
Planned Differentiation Engine (Phase 7)
Three-level language rewriting (Access / Standard / Extension) via Claude API. Language rewriting governed by ml_language_standards.md and differentiation_strategies.md. Number simplification deterministic.
Planned Full Generation Pipeline
Lesson JSON in → complete slide deck + card sorts + Amplify CL scripts out. Scraper, schema, slide system, and card sort system are the building blocks. Generation layer (JSON → HTML) is the remaining major component.

Part 3 — Research Foundations

Key citations supporting each MASL design decision. Page numbers should be verified against institutional library access before formal submission.

As part of his graduate work at Temple University, Bennett developed DeepNote — an interactive HTML study tool generated from source PDFs. Each DeepNote processes an article into ten navigable sections: Tone · Visual Metaphor · Summary · Glossary · Program Connection · Key Quotes · References · Quiz · Card Sort · Reflect. The tool is designed to surface what matters in a source quickly, without losing the argument's structure or nuance.

Eight core sources for this project have been processed as DeepNotes. The standard Course Connection section — which normally maps a reading to syllabus objectives and upcoming assignments — has been adapted here as Program Connection: each DeepNote instead asks how the paper grounds a specific MASL design decision, which activity type or student population it addresses, and what it would mean if the claim turned out to be wrong.

Eight sources — interactive study guides

Arcavi (1994) ↗What fluent symbol reading actually looks like
Sfard (2007) ↗Mathematical thinking is a form of communication
Ke & Newton (2024) ↗Language-supported worked examples for English learners
Barko-Alva & Chang-Bacon (2023) ↗When sentence frames restrict instead of scaffold
Swan (2006) ↗Card sorts produce more math talk than any other task type
Zwiers et al. (2017) ↗Eight routines for math language development
Mezirow (1997) ↗Disorienting dilemma → reflection → transformation
Moschkovich (2015) ↗MLs need the full communicative register, not just vocabulary

Notation Language and Symbol Sense

Card Sorts — Symbol, Speech, and Meaning
Moschkovich establishes that MLs need explicit exposure to the full register of mathematical communication — notation, oral language, and meaning-making: "Learning to communicate mathematically is not a matter of learning vocabulary… students also need to learn to use mathematical discourse." The three-card structure applies the Frayer model's principle that concept acquisition requires both examples and definitional language presented simultaneously. 📖 DeepNote — Moschkovich (2015) ↗
Moschkovich, J. N. (2015). Academic literacy in mathematics for English learners. Journal of Mathematical Behavior, 40, 43–62. · Frayer, D. A., Frederick, W. C., & Klausmeier, H. J. (1969). A schema for testing the level of concept mastery. University of Wisconsin.
Language Writing Standards — Comprehensible Input
The ML-first writing rules operationalize Krashen's comprehensible input hypothesis. Zwiers extends this to academic contexts: reduced linguistic complexity frees working memory for content reasoning rather than text decoding.
Krashen, S. D. (1982). Principles and practice in second language acquisition. Pergamon.
📖 DeepNote — Zwiers et al. (2017) ↗
Zwiers, J., Dieckmann, J., Rutherford-Quach, S., Daro, V., Skarin, R., Weiss, S., & Malamut, J. (2017). Principles for the design of mathematics curricula: Promoting language and content development. Stanford University, UL/SCALE.
Sentence Frames — Scaffolded Output
Gibbons' "mode bridging" — output scaffolds that let students produce language at a register above what they could generate independently. Swain's output hypothesis: production, not just comprehension, drives language acquisition. Barko-Alva & Chang-Bacon's overframing critique is directly engaged: MASL frames never presuppose the answer, and frames are evaluated as a collection across the slide — not individually — to ensure they leave the content open.
Swain, M. (1985, 1995, 2000). Output hypothesis / collaborative dialogue. In multiple works including Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 235–253). Newbury House.
📖 DeepNote — Barko-Alva & Chang-Bacon (2023) ↗
Barko-Alva, K., & Chang-Bacon, C. K. (2023). Over-framing: Interrogating sentence frames as pedagogical support vs. language restriction. Language, Culture and Curriculum, 36(4), 422–438.
How To Reference Slides — Worked Examples and Cognitive Load
Atkinson et al.'s review of worked example research: studying worked examples reduces cognitive load during initial skill acquisition. Barbieri et al. (2023) meta-analysis: g = 0.48 across worked example studies. AlgebraByExample RCT: +7 percentage points on state assessment; +10 for bottom half.
Barbieri, C.A. et al. (2023). Educational Psychology Review. · Booth, J.L., Barbieri, C., et al. AlgebraByExample. IES-funded RCT.
Partner Card Sort — Cooperative Learning and Language Production
Swan (2006): card sorts and matching activities produce more mathematical discussion than any other task type. Sfard's commognitive framework: mathematical thinking develops through communication — peer dialogue during the card sort is not just social, it is cognitive. Roseth, Johnson & Johnson (2008): cooperative learning d = 0.46–0.65. Bowman-Perrott et al. (2016): ELL peer tutoring produces significant language production gains. 📖 DeepNote — Swan (2006) ↗
Swan, M. (2006). Collaborative learning in mathematics: A challenge to our beliefs and practices. National Institute of Adult Continuing Education (NIACE) / National Research and Development Centre (NRDC).
📖 DeepNote — Sfard (2007) ↗
Sfard, A. (2007). When the rules of discourse change, but nobody tells you: Making sense of mathematics learning from a commognitive standpoint. Journal of the Learning Sciences, 16(4), 565–613.
Roseth, C. J., Johnson, D. W., & Johnson, R. T. (2008). Promoting early adolescents' achievement and peer relationships: The effects of cooperative, competitive, and individualistic goal structures. Psychological Bulletin, 134(2), 223–246.
Amplify Feedback — Formative Assessment
Hattie & Timperley: feedback that redirects to the task produces d = 0.79. Black & Wiliam: feedback must prompt further thinking rather than close the task.
Black, P. & Wiliam, D. (1998). Assessment in Education, 5(1), 7–74.

Math as Culture Layer

Ethnomathematics — Cultural Dignity Rationale
D'Ambrosio's foundational claim: mathematics is a cultural product — every civilization develops its own mathematical practices. Instruction that ignores this implicitly devalues students' non-Western mathematical heritage. Every MASL reading that names a specific civilization and mathematician instantiates this framework in practice.
D'Ambrosio, U. (1985). Ethnomathematics and its place in the history and pedagogy of mathematics. For the Learning of Mathematics, 5(1), 44–48. · Bishop, A.J. (1988). Mathematics education in its cultural context. Educational Studies in Mathematics, 19(2), 179–191.
History of Mathematics — Retention and Motivation
Lim & Chapman (2015): quasi-experiment, n=103 Grade 11 students. History-integrated lessons produced significant positive effects on mathematics achievement at immediate posttest and at 4-month and 1-year retention tests. The long-term retention finding is directly relevant to the MASL rationale for unit-opening historical engagement.
Lim, S.Y. & Chapman, E. (2015). Educational Studies in Mathematics, 90(2), 189–212.
Culturally Responsive Mathematics Pedagogy — Equity Warrant
Ladson-Billings (1995): three conditions for culturally relevant pedagogy — academic success, cultural competence, critical consciousness. Gutiérrez (2018): conventional school mathematics dehumanizes students by presenting itself as authorless and culturally European by default. Making non-Western contributions visible is a rehumanizing act. For Philadelphia public school students — many of whom are students of color and English language learners — the claim that their heritage has no relationship to mathematics is not just wrong; it is actively harmful.
Ladson-Billings, G. (1995). Journal of Negro Education, 64(4), 159–160. · Gutiérrez, R. (2018). In Goffney & Gutiérrez (Eds.), Rehumanizing Mathematics. NCTM.
Short Text + Structured Writing — The Bridge Must Have Two Spans
Moje et al. (2004): five-year study with Spanish-English bilingual middle schoolers. Cultural bridge between students' everyday funds of knowledge and disciplinary discourse improved literacy and engagement. Critical finding: both the reading AND the structured follow-up writing are necessary — neither alone completed the bridge. This is the strongest single warrant for pairing reading and prompt as one beat.
Moje, E.B. et al. (2004). Working toward third space in content area literacy. Reading Research Quarterly, 39(1), 38–70.
Reflective Writing — Metacognitive Trigger Prompts
Bangert-Drowns, Hurley & Wilkinson (2004): d = 0.26 across 48 writing-to-learn studies; metacognitive-trigger prompts outperformed summary and note-taking prompts by a significant margin. Graham, Kiuhara & MacKay (2020): g = 0.30 across 56 experiments — writing about content consistently improves learning. The MASL reflection prompt structure is designed as a metacognitive trigger, not a comprehension check: it asks "what does this change about how you see this notation?" — not "what did you read?"
Bangert-Drowns, R.L. et al. (2004). Review of Educational Research, 74(1), 29–58. · Graham, S. et al. (2020). Review of Educational Research, 90(2).
Transformative Learning — Theoretical Frame for Reflection
Mezirow's transformative learning theory: a "disorienting dilemma" — an experience that contradicts the learner's existing frame of reference — can trigger critical reflection and ultimately a transformation in perspective. The historical reading is a mild disorienting dilemma: the notation you use today was not inevitable; it was created by a specific person, in a specific place, to solve a specific problem. The reflection prompt completes the Mezirow cycle by asking the student to articulate what shifted. 📖 DeepNote — Mezirow (1997) ↗
Mezirow, J. (1997). Transformative learning: Theory to practice. New Directions for Adult and Continuing Education, 74, 5–12.
Expressive Writing and Math Anxiety
Park, Ramirez & Beilock (2014): 80 participants, experimental group wrote freely for 7 minutes before a math test. The performance gap between high-math-anxious and low-math-anxious students was substantially reduced in the writing condition. Brief pre-task writing that externalizes anxious thoughts frees working memory for the mathematics itself.
Park, D., Ramirez, G. & Beilock, S.L. (2014). Journal of Experimental Psychology: Applied, 20(2).
Prediction Quiz as Pre-Reading Activation — Format Choice for MLs
MASL uses a prediction quiz rather than an anticipation guide as the pre-reading activation format for its historical readings. The anticipation guide's mechanism — surface a belief, encounter dissonance, revise — requires students to hold a relevant prior belief; many multilingual learners, whose prior schooling may not have included the history of mathematics, have no such belief to examine. A prediction quiz asks students to speculate, which requires only curiosity and is universally accessible regardless of prior schooling context. Three research frameworks ground the choice: schema theory (Anderson & Pearson, 1984) establishes that activating prior knowledge before reading improves comprehension, and a prediction creates a schema hook even without prior knowledge by generating a hypothesis the reading can then evaluate; the SIOP Building Background component (Echevarria, Vogt & Short, 2017) explicitly names pre-reading prediction as high-leverage for English learners; and Krashen's Affective Filter Hypothesis (1982) supports the format's lower anxiety profile — speculation does not put a student's prior schooling on display. The multiple-choice format also reduces language production demands at the entry point, which is a genuine cognitive load advantage for students with lower English proficiency. It should be noted honestly that no head-to-head RCT comparing prediction quizzes to anticipation guides with multilingual learners in secondary mathematics exists in the literature; the format is theoretically grounded, not empirically validated through direct comparison.
Anderson, R.C., & Pearson, P.D. (1984). A schema-theoretic view of basic processes in reading comprehension. In P.D. Pearson (Ed.), Handbook of Reading Research (pp. 255–291). Longman. · Echevarria, J., Vogt, M., & Short, D. (2017). Making content comprehensible for English learners: The SIOP model (5th ed.). Pearson. · Krashen, S.D. (1982). Principles and practice in second language acquisition. Pergamon.