Language courses sell confidence along with content. "Learn in your own style." "Just believe you can grow." "Highlight the key points and re-read." Each piece of advice sounds reasonable, until you examine the data. Here are five widely repeated beliefs and what controlled experiments and meta-analyses actually show.
Where Myths Come From: One Study Is Not a Fact
Educational research has a hierarchy of evidence. At the top: systematic reviews and meta-analyses aggregating dozens or hundreds of experiments. Below that: a single randomised controlled trial. Lower still: an observational study with one group and no control. Most instructional myths grow from that bottom category. A promising result gets published, a journalist writes "proven," and the refutation appears quietly. Publication bias compounds the problem: according to Li (2010), positive results reach print far more often than null or negative ones, systematically inflating estimates of a method's usefulness.
Myth 1: Learning Styles (Visual / Auditory / Kinaesthetic)
The idea is intuitive: some learners absorb material better through images, others through sound, others through movement. Hundreds of courses and curricula are built around it. The trouble is that the meshing hypothesis (matching the delivery format to the learner's preferred style) has never been confirmed in a rigorous experiment.
Pashler, McDaniel, Rohrer, and Bjork (2008) conducted a systematic review and found no reliable evidence that adapting instruction to a learner's style produces better outcomes. The required experimental design (different learning-style groups assigned to different formats, then all groups given the same test) was not met by any study they identified. DOI: 10.1111/j.1539-6053.2009.01038.x.
In practice: a "find your learning style" quiz does not predict language acquisition speed. Mixed-format instruction (text + audio + interactive) outperforms single-channel delivery for most learners: not because it covers all styles, but because it activates multiple memory pathways simultaneously.
Myth 2: Growth Mindset Is a Universal Academic Booster
Since Carol Dweck's 2006 book, the idea that believing your abilities are developable leads to better results has inspired thousands of school programmes. The 2018 evidence paints a more sobering picture.
Sisk, Burgoyne, and colleagues (2018) synthesised 43 studies with more than 57,000 participants. The overall effect on academic achievement was weak (r ≈ 0.10). Brief mindset interventions, such as telling students their brain can grow, did not move grades without accompanying academic support. A meaningful (though still modest) benefit did emerge for at-risk students (first-generation college students, low-income groups). DOI: 10.1177/0956797617739704.
Myth 3: Highlighting and Re-Reading Consolidate Memory
Highlighter and repeated reading are the two most common student strategies. Dunlosky and colleagues (2013) evaluated 10 learning techniques against criteria of study quality, age range, and real-world applicability. Highlighting, re-reading, the keyword mnemonic, and imagery techniques all received a rating of low utility.
Re-reading produces a fluency illusion: the text becomes familiar, and the brain mistakes easy recognition for genuine understanding. Retrieval practice, actively recalling information from memory, works far better. Rowland (2014) documented a retrieval practice effect of d = 0.87 across 47 studies and 6,801 participants.
Myth 4: An App Alone Can Replace Structured Language Learning
Loewen and colleagues (2019) published a correlational Duolingo study often cited as proof of effectiveness. Sample size: 9 participants, no control group. DOI: 10.1017/S0958344019000065. Smith, Jiang, and Peters (2024) ran a larger study: 48 participants, single group, approximately 27 hours of use. DOI: 10.64152/10125/73595. Without a control group it is impossible to separate the app's effect from general exposure or participant motivation.
Apps apply real mechanisms, such as retrieval practice and spaced repetition, and are useful for maintaining vocabulary. But the FSI estimates roughly 600 guided hours to reach B1 for English speakers. No existing study demonstrates that an app alone covers that volume.
Myth 5: Interleaving Always Beats Blocked Practice
Interleaving, or mixing different categories during study, does work, but not universally. Brunmair and Richter's (2019) meta-analysis identified a critical moderator: category similarity. When problem types are similar (e.g. closely related maths problems), interleaving helps. When material is heterogeneous (isolated vocabulary items, descriptive texts), the interleaving advantage shrinks or disappears. DOI: 10.1037/bul0000209.
What Works Instead: Evidence-Based Methods
| Method | Effect size | Source |
|---|---|---|
| Retrieval practice | d = 0.87 (47 studies, N = 6,801) | Rowland, 2014 |
| Spaced repetition | medium–large (48 experiments, N = 3,411) | Kim & Webb, 2022 |
| Post-learning sleep | g = 0.50 (25 studies, N = 1,396) | Schimke et al., 2021 |
| Task-based language teaching (TBLT) | d = 0.93 (52 studies) | Bryfonski & McKay, 2017 |
Using Plonsky and Oswald's (2014) SLA calibration (d ≈ 0.4 small, d ≈ 0.7 medium, d ≈ 1.0 large), both retrieval practice and TBLT fall in the medium-to-large range. DOI: 10.1111/lang.12479.
How Lexy Avoids the Myths
Lexy relies strictly on evidence-backed methods, structuring each lesson across 49 languages into seven complementary stages:
- 1. Vocabulary: interactive flashcards and a matching quiz with active retrieval practice (Rowland d = 0.87) rather than passive reading.
- 2. Warm-up: a short written chat with the assistant to prime communicative schemas.
- 3. Lesson: a 25-minute live voice dialogue with the avatar without typing, solving communicative tasks (TBLT d = 0.93) without error interruption in lessons 1–15 to keep the affective filter low.
- 4. Practice: written phrase exercises, listening, and keyboard-based tests to consolidate grammar and prevent vocal fatigue.
- 5. Dialogue: two-voice situational listening to develop natural comprehension.
- 6. Story: reading story with sentence-by-sentence translation and transliteration.
- 7. Memory Flashcards: spaced repetition of vocabulary from the prior lesson, starting from lesson 2.
Lexy rejects pseudo-scientific learning-style quizzes (multimodal exposure helps all learners) and eliminates stressful live-speech corrections.
Key Takeaways
- Learning styles (visual/auditory/kinaesthetic) have not been confirmed by any study with the required experimental design (Pashler et al., 2008).
- Growth mindset shows a weak average effect (r ≈ 0.10 across 57,000+ participants); meaningful only for at-risk groups (Sisk et al., 2018).
- Highlighting and re-reading rate as low utility; retrieval practice at d = 0.87 and TBLT at d = 0.93 are reliably effective (Rowland, 2014; Bryfonski & McKay, 2017).
- App effectiveness studies with 9 or 48 participants and no control groups cannot support broad claims about language learning outcomes (Loewen et al., 2019; Smith et al., 2024).
Key Points on Language Learning Myths
- Learning styles (visual/auditory/kinaesthetic) have no confirmed experimental support: not a single study with the correct design shows an effect (Pashler et al., 2008).
- Growth mindset effect is weak (r ≈ 0.10, 57,000+ participants); retrieval practice achieves d = 0.87 and TBLT d = 0.93 in meta-analyses.
- Major Duolingo studies covered 9 and 48 participants with no control groups, insufficient to establish effectiveness.
Test Your Knowledge: Article Quiz
90-second quiz: check whether you understood the key themes of this article. 5 quick questions. If you make a mistake, we explain it right away.
A course built on evidence
Lexy's 16 lessons use retrieval practice and TBLT, methods with documented effect sizes. First lesson free.
Start Free Lesson →