I want to share an idea about an online test (generated by a program from comments of redditors). The point of the test is to try to identify "speech patterns" of people you never knew.
This test gives you N blocks of quotes. Quotes in each block are from the same person. You need to guess what blocks don't correspond to a single person. You're not forced to guess everything in the test, just avoid making mistakes. Here's an illustration of this format: (4 blocks, 10 quotes in each)
This test is supposed to test if there're universal patterns in speech that don't have anything to do with "usual" speech metrics (such as word frequencies, word length, specific pairs of words and other... see Stylomerty). If universal patterns exist you can confuse two people with the same speech pattern, but can't confuse two people with different speech patterns, that's why the test's question is formulated the way it is. If universal patterns exist, it also would be a discovery.
Imagine looking at random quotes of people you don't know (funny or sad, thoughtful or just important words) and trying to understand something core about their speech... wouldn't it be interesting/valuable experience?
If those generated speech tests are solvable it won't immediately prove universal speech patterns scientifically. Because, for example, personal information may get into the quotes and serve as hints. But generated tests can kind of "morally prove" the patterns (making everyone convinced that the patterns 90% likely indeed exist) and spark great interest in further investigation.
And I think that the discovery of speech patterns can change the world.
Because some people need an emotional trigger to start fighting for change, to overcome fear.
And I think a discovery like this can be a strong trigger. Or create many emotional triggers for a person, making a person think about other people x2 or x10 times more than they thought before. So... does it hurt to try?
I believe that speech patterns exist because the speech of different people feels drastically different to me. But I don't have the resources to test the idea/to train myself. That's why so far I've only been writing analyzes of dialogue in fictional works to try to get people interested in the topic. For me this is urgent given what happens in the world. Sorry that my idea sounds like an old sci-fi short story.
This test gives you N blocks of quotes. Quotes in each block are from the same person. You need to guess what blocks don't correspond to a single person. You're not forced to guess everything in the test, just avoid making mistakes. Here's an illustration of this format: (4 blocks, 10 quotes in each)
- Block A: (quote 1), (quote 2) ... (quote 10) from person X
- Block B: (quote 1), (quote 2) ... (quote 10) from person Y
- Block C: (quote 1), (quote 2) ... (quote 10) from person Y
- Block D: (quote 1), (quote 2) ... (quote 10) from person Z
This test is supposed to test if there're universal patterns in speech that don't have anything to do with "usual" speech metrics (such as word frequencies, word length, specific pairs of words and other... see Stylomerty). If universal patterns exist you can confuse two people with the same speech pattern, but can't confuse two people with different speech patterns, that's why the test's question is formulated the way it is. If universal patterns exist, it also would be a discovery.
Imagine looking at random quotes of people you don't know (funny or sad, thoughtful or just important words) and trying to understand something core about their speech... wouldn't it be interesting/valuable experience?
If those generated speech tests are solvable it won't immediately prove universal speech patterns scientifically. Because, for example, personal information may get into the quotes and serve as hints. But generated tests can kind of "morally prove" the patterns (making everyone convinced that the patterns 90% likely indeed exist) and spark great interest in further investigation.
And I think that the discovery of speech patterns can change the world.
Because some people need an emotional trigger to start fighting for change, to overcome fear.
And I think a discovery like this can be a strong trigger. Or create many emotional triggers for a person, making a person think about other people x2 or x10 times more than they thought before. So... does it hurt to try?
I believe that speech patterns exist because the speech of different people feels drastically different to me. But I don't have the resources to test the idea/to train myself. That's why so far I've only been writing analyzes of dialogue in fictional works to try to get people interested in the topic. For me this is urgent given what happens in the world. Sorry that my idea sounds like an old sci-fi short story.
