Four evidence-based methods. 50+ years of cognitive science research. 24 peer-reviewed citations. Here's why it works.
Most language apps rely on one method: repetition. MyLinguist combines four, each targeting a different memory system. When they work together, vocabulary doesn't just stick for a test -- it sticks for life.
d = 0.50 · Atkinson & Raugh, 1975
93% vs 13% recall · Bower & Clark, 1969
d = 0.85 · Cepeda et al., 2006
d = 0.74 · Roediger & Karpicke, 2006
Sound-alike bridging turns arbitrary foreign sounds into memorable connections.
In 1975, Stanford psychologists Richard Atkinson and Michael Raugh published a study that changed vocabulary instruction. Students learning Russian vocabulary with phonetic keyword mnemonics scored 88% on the final test, compared to 28% for the free-study control. They learned 120 new words in three days.
The method works through a two-stage process. First, the foreign word is connected to a familiar sound (the "keyword"). Then that keyword is linked to the meaning through a vivid mental image. This converts an arbitrary association into a chain of meaningful connections, dramatically reducing what linguist Paul Nation calls the "learning burden."
A 2021 meta-analysis by Donoghue and Hattie across 242 studies found the keyword method produces a Cohen's d of 0.50. For its specific intended use -- foreign vocabulary acquisition -- the evidence is strongest. When combined with spaced retrieval practice (as we do), the long-term retention concerns are directly addressed.
Atkinson & Raugh (1975), Journal of Experimental Psychology · Raugh & Atkinson (1975), Journal of Educational Psychology · Donoghue & Hattie (2021), Frontiers in Education
Stories encode in the same memory system as life experiences. Flashcards don't.
In a landmark 1969 study, Bower and Clark had subjects learn 12 lists of 10 nouns. Both groups achieved perfect immediate recall. But on delayed recall across all lists, the narrative group remembered 93% compared to 13% for controls -- a seven-fold advantage.
Endel Tulving's foundational work distinguished between episodic memory (memory for personally experienced events) and semantic memory (memory for facts). Rote vocabulary study targets semantic memory -- the weakest pathway for new associations. Story-based encoding targets episodic memory, the same system that records life experiences. That's why you remember riding in a cab more vividly than a page in a textbook.
The absurdity matters too, but not for the reason you'd think. McDaniel and Einstein (1986) showed that it's distinctiveness, not bizarreness itself, that drives the memory advantage. A bizarre story stands out against the background of your normal day -- which is exactly how our stories arrive: 3 vivid scenes among your regular messages.
Bower & Clark (1969), Psychonomic Science · Tulving (1972), Organization of Memory · Craik & Lockhart (1972), Journal of Verbal Learning and Verbal Behavior · McDaniel & Einstein (1986), Journal of Experimental Psychology
The single most effective learning technique known to cognitive science.
The spacing effect was discovered by Hermann Ebbinghaus in 1885 and has been replicated thousands of times since. Cepeda et al. (2006) synthesized 839 assessments across 317 experiments and confirmed: distributing practice over time consistently produces better retention than cramming the same amount of practice into one session.
The 2021 Donoghue and Hattie meta-analysis, drawing on 152,952 participants, found distributed practice produced the highest effect size of all 10 learning techniques studied: Cohen's d = 0.85. Dunlosky et al. (2013) gave it one of only two "high utility" ratings in their comprehensive review -- effective across ages, ability levels, and content domains.
Each spaced encounter happens in a slightly different context, creating multiple retrieval routes to the same memory. The partial forgetting between sessions makes each recall more effortful -- and effortful retrieval strengthens memory more than easy retrieval. This is what Robert Bjork (1994) calls "desirable difficulties."
Ebbinghaus (1885) · Cepeda et al. (2006), Psychological Bulletin · Cepeda et al. (2008), Psychological Science · Dunlosky et al. (2013), Psychological Science in the Public Interest · Donoghue & Hattie (2021), Frontiers in Education
Retrieval practice beats re-study every time. The research isn't even close.
Roediger and Karpicke's 2006 study revealed a striking pattern. Students who studied material four times recalled 83% after 5 minutes -- but only 40% after one week. Students who studied once and tested three times recalled 71% initially, but held 61% after one week. The ranking completely inverted.
In a follow-up published in Science, Karpicke and Roediger (2008) found that students who practiced retrieving Swahili vocabulary recalled ~80% after one week, compared to ~33% for those who only re-studied. Their conclusion was unambiguous: "Repeated studying after learning had no effect on delayed recall, but repeated testing produced a large positive effect."
There's also a metacognitive trap: people systematically overestimate the value of re-reading and underestimate the value of testing. The students who re-read predicted the best performance -- and delivered the worst. The effort of retrieval feels less productive than the fluency of re-reading, but it's far more effective.
Roediger & Karpicke (2006), Psychological Science · Karpicke & Roediger (2008), Science · MacLeod et al. (2010), Journal of Experimental Psychology · Donoghue & Hattie (2021), Frontiers in Education
From Donoghue & Hattie's 2021 meta-analysis: 242 studies, 169,179 participants.
| Technique | Effect Size (d) | Rating | Used by MyLinguist |
|---|---|---|---|
| Distributed Practice (Spacing) | 0.85 | High | Yes |
| Practice Testing (Active Recall) | 0.74 | High | Yes |
| Elaborative Interrogation | 0.56 | Moderate | |
| Imagery Use (Dual Coding) | 0.56 | Moderate | Yes |
| Self-Explanation | 0.54 | Moderate | |
| Keyword Mnemonic | 0.50 | Low* | Yes |
| Re-reading | 0.47 | Low | |
| Interleaved Practice | 0.47 | Low | |
| Underlining | 0.44 | Low | |
| Summarization | 0.44 | Low |
*The keyword mnemonic's "Low" general rating is specific to all-purpose learning. For its intended use -- foreign vocabulary acquisition -- the evidence is substantially stronger. When combined with spaced retrieval, the long-term retention concern is directly addressed.
Each of these four methods is individually well-supported by decades of peer-reviewed research. The specific combination -- 3 story-anchored words per day via messaging with conversational practice -- is our application of these established principles.
No single study has tested this exact four-method combination as a bundle. That's normal for applied products: the science validates each ingredient, and the combination is sound applied cognitive science. We believe in being transparent about what the research shows and what is our own informed engineering.
3 words a day. Stories you can't forget. Science that actually works.