Thanks a lot for your contributions, which trigger a few questions on my side.
@engine: If it's a bot, wouldn't that traffic be "direct" traffic? I can see that a bot scrapes search results, but why would that bot do clicks on the search results? Also, I would expect that somebody would set a bot on money keywords and not long-tail keywords, especially ones that include spelling errors.
@lucy24: I don't have access to the server logs, so I don't have IP addresses, but in GA all the searches come from the same town and it's 100% returning traffic, and they are rather small towns, which increases the likelyhood that it's one single computer.
What you say could make sense, but even if a browser reopens automatically with an open tab that the user has forgotten about, wouldn't GA interpret this as bookmarked access and therefore direct traffic?
Also one would need to assume that the tab remains open for a loooong time.
"humanitarian environment economy" gets traffic of that kind since May 8th
"cooperation in development and humanitarin action" gets traffic since May 26th
Finally, this would mean that the users close and reopen their browsers several times a day.
@vik_c: I think the classroom idea is rather improbable as the bounce rate is 100% (or almost) and students would sometimes explore the website I guess. Also, the terms don't seem to be natural terms for such assignments, expect maybe the ones including internship.