So, as a quick example of how a card-sort exercise might be constructed in this case…
You’d sit down a participant and spread out a shuffled deck of cards on a table, and ask them to put the cards in piles based on how they would think best to catagorise them, and then you ask them why they made these decisions. And you can also gently prompt them to vocalise their thinking as they work.
Our deck could perhaps be:
http://email.google
http://jones.baker
http://email.bbc
http://baker.anderson
http://blog.jones
http://jones.blog
http://google.com
http://bbc.com
http://baker.anderson
http://google.blog
http://blog.google
http://jones.anderson
http://anderson.baker
http://baker.jones
http://anderson.jones
Notes:
- I’d include the protocol in there on each, so folk can contextualise it, “ah, so these are websites”.
- Include some familiar names, and structures that they already find on the web
- Also include a few sets that are subtly ambiguous, i.e. they don’t have with a service like descriptor, and each could be interpreted as a surname, so it’s not too leading.
That’s just a quick thought about how a test might be constructed.