How we choose and verify language resources
Editorial standards for Study New Languages — verification, exclusions, dating, and how to read a resource listing honestly.
The job of a listing
A resource row should help you decide in thirty seconds: what it is, who it is for, what it costs, and whether the link still works. It is not a ranking bought by advertisers.
Verification habits
- Click-through checks on a rolling basis; dead links get flagged or removed.
- Type labels (course_free, media, reference…) so you can filter by job.
- Age-unsafe and scammy “fluency in 7 days” junk stays out.
- Community-governed languages prioritize official community programs over random apps.
Methodology short page: how we choose resources. This essay is the fuller trust document.
How you should read our lists
Start with one free course + one listening habit + one reference. Ignore the rest until those three are alive. Constructed languages are hobby paths, not studio certifications. Revitalization contexts deserve community authority first.
Corrections
Found a dead link or mislabeled resource? Contact or contribute via contribute. Dated honesty beats a bloated museum of URLs.
How a listing earns its row
- Click-through still works (or it gets flagged).
- Type label matches the job (course, media, reference…).
- Age-unsafe and scammy “fluency in 7 days” junk stays out.
- Community-governed languages prioritize official programs.
Verified dates mean a human checked. Blank dates mean not yet re-checked — better than a fake stamp. Corrections welcome via contact / contribute.
Affiliate links, when used, are disclosed. Free directories and tools stay free regardless — see disclosure.
Self-score rubric (every Friday)
- Task (0–2) — answered the brief completely
- Register (0–2) — formality matches the situation
- Collocation / connectors (0–2) — natural glue, not calques
- Control (0–2) — forms stable under time pressure
Aim for 6/8 before increasing difficulty. Below 4/8: repeat the same brief once — do not add new apps.
Related: tools · guides · academic banks · teachers sequence · B2→C1 emails.
Worked cycle 1
Complete one media or reading input, one production task with model compare, and one register sprint. Write down a single phrase you will reuse in a real message this week. If you skip the reuse step, the cycle does not count.
Time-box: twelve minutes to draft, three minutes to edit for register only. Speak the final version once. If speech and writing disagree, simplify the written text until both match.
Worked cycle 2
Complete one media or reading input, one production task with model compare, and one register sprint. Write down a single phrase you will reuse in a real message this week. If you skip the reuse step, the cycle does not count.
Time-box: twelve minutes to draft, three minutes to edit for register only. Speak the final version once. If speech and writing disagree, simplify the written text until both match.
Worked cycle 3
Complete one media or reading input, one production task with model compare, and one register sprint. Write down a single phrase you will reuse in a real message this week. If you skip the reuse step, the cycle does not count.
Time-box: twelve minutes to draft, three minutes to edit for register only. Speak the final version once. If speech and writing disagree, simplify the written text until both match.
Worked cycle 4
Complete one media or reading input, one production task with model compare, and one register sprint. Write down a single phrase you will reuse in a real message this week. If you skip the reuse step, the cycle does not count.
Time-box: twelve minutes to draft, three minutes to edit for register only. Speak the final version once. If speech and writing disagree, simplify the written text until both match.
Worked cycle 5
Complete one media or reading input, one production task with model compare, and one register sprint. Write down a single phrase you will reuse in a real message this week. If you skip the reuse step, the cycle does not count.
Time-box: twelve minutes to draft, three minutes to edit for register only. Speak the final version once. If speech and writing disagree, simplify the written text until both match.
What we will not list
Age-unsafe content, malware-y download farms, and “fluent in 7 days” fantasy. Constructed languages are labeled as hobby paths. Revitalization contexts prioritize community authority.
How to read a row
Name, job type, cost, verification honesty. Start with one course + one listen + one reference.
Closing practice contract
For the next seven days on resource verification, complete three finished loops only: input with retell, production with model compare, and a fifteen-minute register or collocation sprint. Write scores in a simple table. If any loop is skipped, restart the seven-day contract rather than stacking guilt. Tools and guides on this site exist to make that loop obvious — not to multiply options.
Cadence
Rolling link checks, type labels, and community-first ordering for revitalization languages. When a beloved app turns hostile or dead, we remove it rather than leave a zombie row.
Reader checklist
Before you trust a listicle elsewhere: does it date checks, label cost, and admit gaps? If not, prefer a shorter honest list.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
Model micro-task for resource-verification: write forty words, speak them, then edit one register feature only (opening, hedge, or closing). Stop when speech and writing agree.
FAQ
- Are affiliate links used?
- When used, they are disclosed. Free directories and tools stay free regardless. See /disclosure/.