The science of Evidence Based Vocabulary
"Research-backed" is the most overused phrase in education software. So instead of claiming it, we'll show our work: every design decision below maps to a specific research tradition, with the canonical citation. If a finding doesn't apply, we don't claim it.
1. Spaced repetition
Hermann Ebbinghaus measured the forgetting curve in 1885 by memorizing nonsense syllables and tracking how quickly he lost them. The curve has been replicated in essentially every memory study since. Without retrieval, roughly 80% of new vocabulary is gone within a week.
The fix is spaced retrieval: bring each word back at expanding intervals, timed to just before it would be forgotten. Cepeda and colleagues' 2006 meta-analysis of 184 studies found spacing improved long-term retention by an average of 11 percentage points over massed practice — a large effect that holds across ages, materials, and time scales.
In Evidence Based Vocabulary: the SM-2 algorithm schedules each word independently. A word answered confidently is pushed further out; a word missed is brought back tomorrow. The schedule adapts per word, per kid.
Ebbinghaus, H. (1885). Über das Gedächtnis. Cepeda, N. et al. (2006). "Distributed practice in verbal recall tasks: A review and quantitative synthesis." Psychological Bulletin.2. Retrieval practice (the testing effect)
A passive review — re-reading a definition, looking at a flashcard's answer side — produces far weaker memory than actively retrieving the answer from memory, even when retrieval is wrong and immediately corrected. Roediger and Karpicke's 2006 paper showed this effect is large enough that a single retrieval attempt outperforms multiple re-reads on a delayed test.
In Evidence Based Vocabulary: every card is a retrieval task. Multiple-choice meaning quizzes, fill-in-the-blank exercises, spelling cards. We never just show the kid a word and a definition.
Roediger, H. L., & Karpicke, J. D. (2006). "Test-enhanced learning: Taking memory tests improves long-term retention." Psychological Science, 17(3), 249–255.3. Dual coding
Allan Paivio's dual-coding theory holds that memory is strengthened when information is encoded both verbally and visually — two retrieval pathways instead of one. Words taught with images are retained substantially better than words taught with definitions alone.
In Evidence Based Vocabulary: every vocabulary word is paired with an image generated to depict that exact word in a memorable scene. Not stock clip-art — a fresh image per word, sized for the meaning.
Paivio, A. (1986). Mental Representations: A Dual Coding Approach. Oxford University Press.4. Morphological awareness
English vocabulary is densely combinatorial. A child who knows that un- means "not," -able means "capable of," and port means "carry" can decompose unportable the first time they meet it. Carlisle's 2010 review and Bowers, Kirby, and Deacon's 2010 meta-analysis both found that explicit morphological instruction produces measurable gains in vocabulary, reading comprehension, and spelling — with the largest effects for younger and struggling readers.
In Evidence Based Vocabulary: kids are introduced to one morpheme at a time, in a deliberate order (transparent prefixes first, derivational suffixes later). When a vocabulary word contains a morpheme they've already learned, the card surfaces the connection: "You know un- means 'not' — so unsteady means…"
Carlisle, J. F. (2010). "Effects of instruction in morphological awareness on literacy achievement." Reading Research Quarterly, 45(4). Bowers, P. N., Kirby, J. R., & Deacon, S. H. (2010). "The effects of morphological instruction on literacy skills: A systematic review." Review of Educational Research, 80(2).5. Depth of word knowledge and polysemy
Knowing a word isn't binary — Nagy and Scott's chapter in the Handbook of Reading Research describes word knowledge as multidimensional: meaning, multiple senses, register, collocations, morphological relatives. A child can "know" bright in the sense of "shining" but be lost when the word means "intelligent" or "vivid."
Beck, McKeown, and Kucan's Bringing Words to Life argues that the highest-leverage vocabulary instruction targets Tier-2 words: high-utility, polysemous words that appear across academic contexts. These are the words a child has nearly understood from context but not quite, and they are precisely the words a good book hands you for free.
In Evidence Based Vocabulary: when a word with multiple senses comes back for review, we don't just re-test the original meaning. We bank additional excerpts, generate sense-extension cards, and quiz the kid on which sense fits which sentence. The aim is depth, not just breadth.
Nagy, W. E., & Scott, J. A. (2000). "Vocabulary processes." In Handbook of Reading Research. Beck, I. L., McKeown, M. G., & Kucan, L. (2002, 2013). Bringing Words to Life: Robust Vocabulary Instruction.6. Words from real reading (Tier-2 vocabulary)
Vocabulary apps that ship a fixed list of "important words" face a structural problem: the list isn't grounded in any particular child's reading. Cunningham and Stanovich's classic What Reading Does for the Mind documents how dramatically reading volume drives vocabulary growth — but only when readers attend to unfamiliar words rather than skipping them. The Tier-2 framework points to the same answer from a different angle: target the words a kid has just almost understood from context.
In Evidence Based Vocabulary: upload the EPUB your kid is reading. We extract the candidate words (Membean-grade Tier-2 vocabulary that actually appears in this chapter), rank by reading level and difficulty, and teach those. The flashcards are tied to a sentence pulled from the book itself.
Cunningham, A. E., & Stanovich, K. E. (1998). "What Reading Does for the Mind." American Educator. Beck, McKeown, & Kucan (2002).7. Phonological encoding
Baddeley's working-memory model identifies the phonological loop as a key buffer for verbal information: a word's sound is part of how it's stored. Words that are seen but never heard are encoded weakly. Kids who meet a word in print without ever hearing it pronounced often guess wrong, and the wrong pronunciation locks in.
In Evidence Based Vocabulary: every card has audio (Free Dictionary API pronunciation) and IPA phonetics. Kids hear the word every time they meet it, before they ever have to choose a meaning.
Baddeley, A. D. (2003). "Working memory and language: an overview." Journal of Communication Disorders, 36(3).8. Adaptive difficulty (desirable difficulty)
Robert Bjork's research on desirable difficulties — the counterintuitive finding that learners retain more from material that's challenging but not crushing — has held up across forty years of replication. Too easy, and there's no encoding. Too hard, and there's no retrieval. The sweet spot is at the edge of ability.
In Evidence Based Vocabulary: each word in our database is rated on real-world frequency and on how often other students answer it correctly. The app picks new words for your kid in the zone where they're getting roughly 70–85% right — challenging, not punishing. If they're struggling, the difficulty drops; if they're flying, it scales up.
Bjork, R. A., & Bjork, E. L. (2011). "Making things hard on yourself, but in a good way: Creating desirable difficulties to enhance learning."What we're not claiming
Evidence Based Vocabulary hasn't been the subject of a peer-reviewed study. We can't tell you "kids using our app retain X% more vocabulary than kids using Method Y." That study would take years and a research team. What we can tell you is that every design decision in the app is grounded in a finding that has been replicated, often dozens of times, across decades of cognitive psychology and reading research.
If a peer-reviewed efficacy trial of Evidence Based Vocabulary itself ever exists, we'll link it here. Until then, the references on this page are what we have.