Research Stage

Reading Fluency Engine

A reading-first learning app in the research stage, focused on helping learners build automatic, confident, sustained reading fluency.

StageResearch and product architecture
FocusReading fluency, automaticity, comprehension, stamina
Initial PlatformNext.js web app under this portfolio

This project is not being presented as a finished application yet. It is a research and product-architecture project that defines what the future app should measure, how it should adapt, and what constraints should guide the build.

The core idea is that reading fluency is the product. The app should not optimize for finishing books, earning points, or completing lessons. It should help the learner practice the easiest useful text that creates real growth right now.

Research Position

What The Research Is Pointing Toward

The current notes define product constraints before implementation starts. These are the principles the app should preserve as the project moves from research into design and engineering.

  • Progress is based on mastery evidence, not age, grade, XP, or lesson completion.
  • Content should sit inside the learner's Zone of Proximal Development: challenging enough to grow, not so difficult that reading becomes frustrating.
  • The learning engine should be deterministic. Given the same learner data, it should make the same instructional decision.
  • AI can support the experience, but it should not own curriculum decisions or be required for the app to function.
  • Children should not need conventional email-and-password accounts to use the app.
  • Privacy and data minimization have to shape the architecture from the start because the product may serve children.

System Model

Three Layers Of The Future App

The architecture separates instructional decisions, structured reading content, and optional AI enhancements so the core learning experience can remain reliable and explainable.

Reading Engine

The deterministic core that estimates ability, tracks mastery, identifies weak skills, chooses the next objective, schedules review, and determines progression.

  • Learner profile
  • Mastery tracking
  • Skill recommendations
  • Review scheduling

Content Library

A structured library of letters, phonemes, graphemes, words, sentences, passages, articles, stories, and books tagged with skill and difficulty metadata.

  • Decoding difficulty
  • Vocabulary difficulty
  • Grammar patterns
  • Phonics tags

AI Layer

An optional enhancement layer that may explain words, generate hints, simplify passages, summarize stories, or help create reviewed metadata.

  • Hints
  • Passage simplification
  • Metadata assistance
  • Learner questions

Findings

Research Themes

These are the main ideas pulled from the current research sources. They describe what the app should optimize for and what should be avoided.

Foundational Reading Components

The research base keeps pointing back to the same foundation: phonemic awareness, phonics, fluency, vocabulary, and comprehension need to work together. The app should not treat one of those skills as the whole reading problem.

  • NICHD
  • National Reading Panel
  • Phonemic awareness
  • Phonics
  • Fluency

Misconceptions To Avoid

The app should not frame the science of reading as a fad, a single program, or phonics-only instruction. Phonics matters, but the product has to connect word-reading skill to vocabulary, language comprehension, and actual reading.

  • Arizona Department of Education
  • NICHD
  • Not one-size-fits-all
  • Not phonics only

Word Memory And Orthographic Mapping

Readers remember written words when spellings, pronunciations, and meanings become connected in memory. That makes phonemic awareness and grapheme-phoneme knowledge product-critical instead of just lesson categories.

  • Ehri
  • Kilpatrick
  • Orthographic mapping
  • Sight word memory

Automatic Decoding Supports Comprehension

When word recognition is slow or effortful, attention is pulled away from meaning. Fluency practice should build accuracy first, then automaticity, expression, and sustained connected-text reading.

  • LaBerge and Samuels
  • Accuracy
  • Automaticity
  • Prosody
  • Connected text

Explicit Sequence With Application

A progress tree should move from simpler sound and alphabetic skills into more complex spelling patterns, morphology, fluency, and comprehension. Each new node should include both isolated practice and application in real text.

  • UFLI Foundations
  • IES Practice Guide
  • Scope and sequence
  • Word work
  • Connected text

Decoding Plus Language Comprehension

The Simple View of Reading is useful product scaffolding because it keeps the app from reducing reading to word attack alone. A learner may need support in decoding, language comprehension, or both.

  • Gough and Tunmer
  • Simple View of Reading
  • Language comprehension
  • Reading comprehension

Deterministic Progress Instead Of Lesson Completion

The app should still be a deterministic learning system. AI may support hints or metadata work, but the instructional path should be based on observable mastery evidence and a clear skill graph.

  • Mastery over completion
  • Skill evidence
  • Review scheduling
  • Explainable decisions

Progress Tree

Draft Skill Sequence

This is the current shape of the progression model. It is not a finished curriculum yet, but it gives the future app a research-aligned path for sequencing practice and measuring growth.

Sound Awareness

  1. Words in sentences
  2. Syllables
  3. Rhyme and onset-rime
  4. Identify phonemes
  5. Blend phonemes
  6. Segment phonemes
  7. Delete or substitute phonemes

Alphabet And Decoding

  1. Letter recognition
  2. Letter sounds
  3. CVC words
  4. Digraphs
  5. Consonant blends
  6. Silent-e
  7. Vowel teams
  8. R-controlled vowels
  9. Complex spellings
  10. Multisyllable decoding

Word Knowledge

  1. High-frequency irregular words
  2. Orthographic mapping
  3. Spelling patterns
  4. Prefixes and suffixes
  5. Roots and morphology

Fluency

  1. Accurate word reading
  2. Phrase reading
  3. Automaticity
  4. Expression and prosody
  5. Sustained connected-text reading

Language Comprehension

  1. Oral vocabulary
  2. Sentence meaning
  3. Grammar and syntax
  4. Background knowledge
  5. Narrative structure
  6. Informational-text structure
  7. Inference
  8. Comprehension monitoring

A progress-tree node should ideally require both skill mastery and application mastery. That prevents a learner from clearing isolated drills without proving they can use the skill during actual reading.

Skill Mastery

The learner can perform the isolated skill accurately and with enough consistency to show the pattern is learned, such as correctly blending 18 of 20 unfamiliar CVC words.

Application Mastery

The learner can apply that same pattern in connected reading, such as reading sentences or a short passage containing the target pattern with appropriate accuracy and support.

Learner Model

Skills The App Should Track

The learner should not be reduced to one generic reading level. The app should preserve a more specific profile so practice can target the actual source of difficulty.

  • Letter recognition
  • Phoneme and grapheme knowledge
  • Blending
  • Short and long vowels
  • Consonant blends and digraphs
  • Sight words
  • Morphology
  • Vocabulary
  • Sentence comprehension
  • Reading speed
  • Reading stamina
  • Confidence

Product Model

Accounts, Classrooms, And Privacy

The account model is part of the learning design. Progress should belong to the learner, while adults and classrooms receive only the access they need.

  • Adult accounts manage access, consent, security, households, classrooms, and dashboards.
  • Learner profiles store reading history, skill-tree progress, assessments, preferences, and mastery evidence.
  • Children can enter through profile selection, PINs, class codes, QR login cards, or classroom device mode instead of requiring email accounts.
  • Teachers and tutors receive scoped instructional access, not guardian ownership.
  • Removing a classroom or ending educator access must not delete learner progress.

Roadmap

Phased MVP Direction

The first build should validate the instructional model before attempting full school or district infrastructure.

Phase 1: Individual and Family

  • Adult account
  • Household
  • Learner profiles
  • Placement assessment
  • Skill tree
  • Parent dashboard

Phase 2: Lightweight Classroom

  • Educator role
  • Classrooms
  • Manual roster
  • Class codes
  • Assignments
  • Class skill map

Phase 3: School Product

  • Organizations
  • School-year transitions
  • Bulk rostering
  • Exports
  • Privacy controls
  • SSO integrations

References

Research Sources Used

These are the external research, curriculum, and plain-language sources behind the current product direction.

Arizona Department of Education / 2024

Welcome to the Science of Reading

Used for guardrails around common misconceptions: the science of reading is not a fad, not a single program, not one-size-fits-all, and not only phonics.

Open source

University of Florida Literacy Institute / 2022

UFLI Foundations

Used as a curriculum sequencing reference for explicit foundational skills instruction, word work, irregular words, and connected text.

Open source

Internal Working Notes

  • Reading Fluency App - Vision & Technical Contextdocs/reading_app/vision_and_technical_context.md
  • Reading App - Accounts, Learner Profiles, Families, and Classroomsdocs/reading_app/reading_app_accounts_classrooms_context.md