What I read
Where the ideas on this site came from, and what I actually took from each one.
⏱ About 7 min read
Almost nothing on this site is my own idea. I only tested them on myself, kept what worked, and dropped what didn't.
This is where they came from. I have noted what I actually took from each source, because a bare list of links is available everywhere.
The original ideas
| Source | What I took from it |
|---|---|
| Stephen Krashen — books free to read on his own site: Principles and Practice in Second Language Acquisition and The Power of Reading | The comprehensible input idea: the language that gets in is the language you understand, not the language you memorise. The second book is why I started reading fiction instead of textbooks. |
| Paul Nation — nearly all of his papers free to download, including a Vietnamese translation of What do you need to know to learn a foreign language? | The four pieces and every number in How many words is enough. The most practical source I found. |
| Michael Lewis — The Lexical Approach (1993). Free summary: Cambridge's Learning language in chunks | The idea that English is assembled from ready-made chunks, not grammar plus vocabulary. Leads straight into Collocations. |
| Refold — an immersion roadmap, free to read | A clear division into stages, and the "don't speak early" idea. I did not follow all of it, but it did more than anything else to reduce my confusion in year one. |
| Antimoon — an old site, ugly interface, real content | The idea of collecting whole sentences instead of isolated words. The two people who wrote it taught themselves English to near-native level and documented it very frankly. |
I don't follow any source completely
There is a part of every source above that I dropped.
Krashen almost entirely rejects deliberate study — I found Anki and a few weeks of grammar genuinely useful. Refold has rules I find too rigid.
If you follow a system to the point of not daring to deviate, you are probably optimising for the system rather than for yourself.
The primary research I cite numbers from
Every article on this site has a Sources for this page section at the end, pointing at the exact research its numbers came from. This is the combined list, all free to download:
| Study | The number it gives |
|---|---|
| Hu & Nation (2000). Unknown vocabulary density and reading comprehension | The 95% and 98% coverage thresholds |
| Nation (2006). How large a vocabulary is needed for reading and listening? | 6,000–7,000 word families for listening, 8,000–9,000 for reading |
| Nation (2007). The four strands | The four pieces frame |
| Nation (1989). Improving speaking fluency | The 4/3/2 exercise |
| Nation. Principles guiding vocabulary learning through extensive reading | About 1,000 word families a year in a classroom |
| Browne (2014). A New General Service List | The 2,800-word NGSL covering over 92% of text |
| Phonological processes in English connected speech (2025) | The five processes that deform sentences in fast speech |
| Common mistakes in pronouncing English consonant clusters: Vietnamese learners (2022) | ~60% of Vietnamese cluster errors are dropping the second consonant |
| A Systematic Review of Research on the use of Shadowing (2025) | The two stages of shadowing |
| Cambridge English. How long does it take to learn a language? | Guided learning hours per CEFR level |
Which parts are research and which are mine
Any number with a citation is research, and you can check it.
Any number without one — hours I studied, cards I deleted, pages I read — is my own experience. It is useful for getting a sense of scale, but it is not evidence of anything.
Word lists and corpora
| Source | For |
|---|---|
| NGSL | The 2,800 most common words, free download. I use it to decide which words deserve a card. |
| Lextutor | Paste in a passage, see what frequency band it sits in. The fastest way to tell whether a book is too hard for you. |
| COCA | An American English corpus. Check whether a phrasing is genuinely common. |
Dictionaries
I go into how to choose in Monolingual dictionaries. These are the specific ones:
- Oxford Learner's Dictionaries — definitions written with a limited vocabulary. The first one to switch to.
- Merriam-Webster Learner's — similar, American-leaning, with more example sentences.
- Longman — marks which words are in the most common bands, handy for telling whether a word is worth learning.
- ozdic — a collocation dictionary. Small, ugly, extremely useful.
- SkELL — no definitions, just real corpus sentences.
- Forvo — hear real people from different regions say a word.
- YouGlish — hear a word or phrase across hundreds of real videos.
Listening and reading sources
My principle is still pick few, go deep. This list is for choosing one, not for bookmarking all of it.
For beginners:
- BBC Learning English — a graded path, with a transcript for every piece.
- VOA Learning English — news read slowly, with a vocabulary capped at about 1,500 words. Transcripts included.
- ELLLO — thousands of short conversations, many accents, transcripts, sorted by level. This is what I use for dictation.
Reading:
- Extensive Reading Foundation — graded reader lists by level. The safest way in.
- Project Gutenberg — free out-of-copyright books. Note that most are old, with harder prose than everyday English.
Intermediate and up: anything you genuinely want to watch. The only requirement is accurate English subtitles, so you can go back over what you missed.
Tools for mining material
| Tool | Role |
|---|---|
| Anki | Spaced repetition. Free on every platform except iPhone. |
| FSRS setup guide | Anki's newer algorithm: fewer reviews for the same retention. Enable it in deck options. |
| Ddict | Look words up on the page you are reading. The tool I use most. |
| Readlang | Read the web and books in the browser, click a word for its meaning, save it. |
| asbplayer | Sync subtitles to video, clip both the sentence and the audio into Anki. Free and open source. |
| Language Reactor | Dual subtitles and in-place lookups on streaming platforms. The free version is enough. |
| Tatoeba | A bank of example sentences with translations; find real sentences containing the word you are learning. |
Finding people to correct you and speak with
- italki — paid tutors by the session; prices vary enormously by teacher.
- Language exchange apps — free, in return for teaching your own language. Quality varies wildly.
- Learner communities on forums — a good place to ask someone to look over a short piece of writing.