"I knew you were in the park."
Translation:Jag visste att du var i parken.
It’s a difficult question and I have no research to back me up, so I will go with my own intuition, and then ask the other mods if someone knows more:
I think it’s more common with certain verbs, like verbs that introduce some sort of reported speech, such as know, say, think, claim etc. This is still more a spoken phenomenon I think, and the att tends to be included in writing, or at least in more formal writing. In speech, to me, these sentences are fine however.
- Han visste det var nåt som saknades. (He knew [that] something was missing)
- Jag tror han kommer i morgon. (I believe [that] he’s coming tomorrow.)
- Jag sa jag var besviken på honom. (I said [that] I was disappointed in him.)
These sounds more dubious, and probably not very common, but don’t sound completely ungrammatical:
- ?Det är intressant han gillar språk så mycket. (It is interesting [that] he likes languages so much.)
- ?De fastställde det rörde sig om något annat. (They confirmed [that] it concerned something else.)
I would never say something like this, I consider this ungrammatical:
- ﹡Det var roligt han ramlade. (It was fun [that] he fell.)
It also seems to me that the longer the main clause is, the more dubious the omission of that becomes, even though they include the same verbs as above:
- ?Han hade under dagen vetat det var nåt som saknades. (He had during the day known [that] something was missing.)
- ?Hur kan du tro han kommer i morgon? (How can you possibly believe [that] he’s coming tomorrow?)
- ?Alla eleverna i klass 5B sa de var besvikna på honom. (All the pupils in class 5B said [that] they were disappointed in him.)
Also not that this is a different phenomenon than the infinitive marker att that was discussed in this thread.
I shall ask around more and look for scientific papers about the phenomenon, and maybe we can write another text about it that’s not based on my personal intuition.
Nice summary! I'd like to add some things, but please note that I don't have any sources here either.
If you can't exclude the att, it's mostly because the det or a similar referrent has been moved to another place in the sentence. You can see this for non-questions by moving the phrases around to a construction where the att is required, but without changing the word order within phrases. Let's compare:
- Han visste [att] det var nåt som saknades.
- ?[Att] det var nåt som saknades, han visste.
- Det var roligt att han ramlade.
- Att han ramlade, det var roligt.
In the first case, we can see that det var nåt som saknades is a direct argument to han visste, and you can hence leave the att out, since it is implied. But in the second case, att has a slightly different conjunctive function and is required.
It's actually basically the same in English:
- He knew that there was something missing.
- He knew there was something missing.
- It was hilarious that he fell.
- ?It was hilarious he fell.
I think that it would probably be easy to find patterns through analysis, if anyone is so inclined. For instance, I would assume that phrases beginning with det tend to require the att.