The blind read-aloud test
Show a blend to somebody who does not know the pair, without explaining it, and record what they say back. This is step two of the review guide and the one thing no name tool can do for you, so this page just makes it easy to run properly.
Step one, on your own
Enter the blend and the two names behind it. The next screen hides the source names so you can hand the device to someone else.
Nothing on this page is sent anywhere. Everything stays in this browser tab.
Why the blind part matters
Once you know a blend, you cannot unknow it. Tayvis looks obvious the moment somebody tells you it is Taylor and Travis, and it looked like nothing at all a second earlier. That gap is invisible to the person who made the blend, and it is exactly the experience every new reader will have.
So the test has to be blind, and it has to come before any explanation. If you say the names first, you have destroyed the only measurement you were trying to take.
How to run it well
- Pick someone who does not know the couple or the pairing. A colleague is ideal.
- Show the blend on its own. No context, no hint, no expectant face.
- Ask them to say it out loud first, then to guess the two names.
- Write down what they actually said, including a wrong guess. The wrong guesses carry more information than the right ones.
- Only reveal the answer afterwards.
Reading the result
Both names recovered is a strong pass. One name recovered usually means one source is contributing too little to be traced, which you can see as a low coverage figure in the structure report. Neither name recovered means the blend is only meaningful to people who already know it, which is fine for a private nickname and a problem for a tag anyone else has to find.
Pronunciation is the other half. If your reader says it differently from how you say it, the spelling is doing something you did not intend, and every future reader will hit the same fork.
Send us the result
The human review study is collecting exactly this data and currently has none. If you run a test, the result is useful to us whether it passed or failed, and failures are the ones people never send.
Common questions
What is a blind name test?
You show a blended name to someone who does not know the two people behind it, without explaining anything, and record how they say it and which names they guess. It tests whether the blend carries its own meaning to a stranger.
Why does the blind test matter more than the structure report?
Because the structure report reads spelling and a real reader hears sound. A blend can score perfectly on coverage and balance and still be unreadable out loud. The only way to find that out is to put it in front of someone.
How many people should I test a blend on?
Two or three is enough to catch an obvious problem. If the first person recovers both names and pronounces it the way you intended, the blend is doing its job. If two people in a row miss the same name, that is a pattern rather than bad luck.