TITLE: How to analyze this sentence - "I have to wait for I don't know how long."
I know the following sentence is correct:
- I have to wait for I don't know how long.
If we shift "for long" part like this (see #2), it will be incorrect. Right?
- I have to wait for how long I don't know.
I think that sentence #2 is incorrect.
While analyzing the correctness of sentence #1, it's said that its constituents are moved out of their natural position. And when they are placed in their natural positions, the sentence will look like the following:
- I don't know how long I have to wait for.
So far it's good. But I have always known that when constituents are moved they are moved entirely, not in part.
It seems that you might be thinking that all three examples are three versions of the same basic sentence. Let us first see if that is true by first parsing example #3, and then later we'll compare its parse to parses of examples #1 and of #2.
Example #3: A parse of it could be:
- I don't know [ [how long]i I have to wait (for) __ i ].
Notice that the word "for" is optional in this case. And there is a gap in the subordinate clause that is co-indexed with the phrase "how long". The phrase "how long" has been fronted in front of its clause, and where the normal order for that clause would be "I have to wait (for) [how long]", roughly speaking.
To understand how to interpret #3, consider the interrogative main clause:
- 3.b. [How long]i do I have to wait (for) __ i ? <-- (a main clause)
which could be embedded as a subordinate interrogative clause:
- 3.c. I don't know [ how long I have to wait (for) ]. <-- (same as OP's #3)
An interpretation for #3.c (and the OP's #3) is: I don't know the answer to the question "How long do I have to wait (for)?"
The embedded question is "How long I have to wait (for)?", and it is the complement for the verb "know". That is, the expression "how long" is merely the fronted element of that embedded question. Notice that the embedded question is minus the auxiliary verb "do" that was used in the main clause version #3.b.
The whole embedded question (i.e. the subordinate interrogative clause) could be fronted:
- 3.d. [ How long I have to wait (for) ] I don't know.
These last two versions (#3.c and #3.d), they both have an embedded question (a subordinate interrogative clause) as the complement for the verb "know", and the lexical verb "know" is the so-called main verb of the sentence.
Example #1: A parse of it could be:
- 1.b. I have m[ to wait l[ for k[ I don't know j [how long] j ]k ]l ]m . <-- (same as OP's #1)
As we see in #1.b (and OP's #1), the head verb of the main clause is the verb "have", and so, it will appear that your example #1 has a different structure than your example #3 which has the head verb "know".
Example #2: A parse of it could be:
- 2.b. I have m[ to wait l[ for k[ j [how long] j I don't know ]k ]l ]m . <-- (same as OP's #2)
As we see in #2.b (and OP's #2), the head verb of the main clause is the verb "have", and so, it will appear that your example #2 has a different structure than your example #3 which has the head verb "know".
It seems that example #2 might be somewhat similar to example #1, in that they seem to have somewhat similar "for" preposition phrases (PPs). The surface difference seems to be that, for example #2, the PP's complement has an element preposed; that is, the element "how long" was fronted.
But it probably depends on what the speaker was trying to say in #2 and #3, as to what would be an appropriate parse or interpretation. And also, the context would probably be needed.
For both examples #1 and #2: The verb "wait" can take as a complement various different types of categories, it can take an infinitival clause, PP, etc.:
I have to wait [for Tom to steal a bus].
I have to wait [for five o'clock].
I have to wait [for when Tom shows up].
Examples #1 and #2 have different structures from that of example #3. That is, #1 and #2 are not versions of #3.
Examples #1 and #2 seem to perhaps be versions of each other, where one has an element preposed (w.r.t. to the structure of the PP's complement); that is, within that PP's complement of example #2, the element "how long" was fronted.
Examples #1 and #2 sound rather informal, and seem to be something that would be found (acceptable?) in speech or narrative fiction. Though, #2 might be okay. (caveat: I haven't looked too deeply into this.)
There might be possibilities where example #1 (and maybe #2) might be seen as having a structure that is somewhat similar to something like "I have to wait for who knows how long" and other structures. (caveat: I haven't looked too deeply into this either.)
ASIDE: As to the sentence:
- I don't know [ I have to wait [for [how long]] ].
Example #4 is probably not acceptable for standard English. It seems that the example is supposed to involve a subordinate interrogative clause, but the interrogative phrase was not fronted within that subordinate interrogative clause as it would have been expected. And so, it appears to be unacceptable.
That example #4 ("I don't know I have to wait for how long") might be found in informal speech, perhaps due to the speaker thinking at the same time as they are speaking. But in edited prose, or "edited" speech, it would probably be "I don't know how long I have to wait", which is your original #3 example.