Learn: the science of spelling
Plain-language guides to the research, written for teachers, parents and curious adults.
What this section is
The research on spelling runs from laboratory memory experiments through classroom trials to large evidence reviews, and most of what it found never reaches the people teaching or learning it. These pages summarise that work in plain language: how spelling gets into memory, what kinds of practice actually move it there, and what digital tools can and cannot add.
They are informational, not promotional. Nothing here describes SpellingIQ's product decisions; the pages teach what is known about the topic, and every factual claim carries a reference you can follow to the source.
How we treat evidence
Not all citations are equal, and these pages say so as they go. A meta-analysis pooling dozens of studies carries more weight than a single experiment; a single experiment carries more than a professional consensus document; and a company's report about its own product is a claim, not a finding. Where a widely quoted number comes from a secondary summary rather than the original paper, or from a study population that may not generalise, the text says that too.
Grading evidence rather than presenting it all as equally settled is a convention rather than our invention, though each body does it its own way: the US Institute of Education Sciences attaches a supporting-evidence rating to each individual recommendation in its What Works Clearinghouse practice guides, and the UK's Education Endowment Foundation rates the evidence behind its literacy recommendations on quality and quantity, consistency, and relevance. Neither uses the exact ordering above. What they share with it is the refusal to let a weak source and a strong one read the same.
The guides
Spelling is not photographic memory. How written words actually get into long-term memory, covering sound-to-letter mapping, self-teaching, the patterns we absorb without noticing, and why English is more systematic than its reputation.
A century of memory research applies directly to spelling: testing yourself beats rereading, spacing beats cramming, mistakes are productive when corrected, and difficulty has a useful middle. What the studies found, and how confident anyone should be.
Software can adapt difficulty, schedule reviews and make practice pleasant, and the evidence behind those features ranges from randomised trials to marketing claims. How to tell which is which, whatever app you are evaluating.
Who this is for
Teachers will find the instructional evidence: what formal spelling teaching achieves, which practice routines have replicated, and where the literature is genuinely unsettled. Parents will find enough background to evaluate any spelling programme or app, including the questions worth asking of one. Adults working on their own spelling will find out why testing yourself on a word is expected to beat reading it again, the retrieval-practice finding (Karpicke, 2017), which is well replicated for verbal material generally, though the studies behind it mostly used prose and word pairs rather than spelling.
Where SpellingIQ stands
The short version is on the philosophy page: evidence first, honest feedback, difficulty treated as a feature rather than a flaw. For the product itself, start with about, play the Daily Five, or take the placement test.
References
- Karpicke (2017). Retrieval-Based Learning: A Decade of Progress. A review of the decade of work following Roediger & Karpicke's 2006 test-enhanced-learning experiment, which it summarises rather than reports
- IES What Works Clearinghouse. Teaching Elementary School Students to Be Effective Writers (practice guide). Institute of Education Sciences. Each recommendation carries its own evidence rating: the labelling convention described above, applied by the body that publishes it
- Education Endowment Foundation. Improving Literacy in Key Stage 2 (2nd ed., 2021). Guidance reports state the strength of evidence behind each recommendation rather than presenting them as equally settled