He created it. Put me in touch with him.
So which is correct and why:
- Put me in touch with whoever/whomever created it.
Your example involves a fused relative construction. In this construction, the antecedent and the relativized element are fused together instead of being expressed separately as in simpler constructions (2005 textbook ASITEG, page 191).
The pronoun who(m)ever is only used either as a relative pronoun in fused relatives or as an interrogative pronoun in the exceptional interrogative construction that functions as exhaustive conditional adjunct ("I shan't be attending the meeting, [whoever takes over as chair]"). -- CGEL page 429.
(Note: There is another issue that has potential to make things even more confusing, and that is when the fused relative is in a "free choice" construction. I'll address that issue near the bottom of this post. At first blush, your example doesn't seem to involve a "free choice" usage.)
In your example, as you have noticed, the relative pronoun "who(m)ever" has to fulfill two different case requirements. The matrix clause requires the relative pronoun to be in accusative case:
- Put me in touch with him.
while the relative clause requires the pronoun to be in nominative case:
- Put me in touch with the person [who created it].
In general, there are the three possibilities:
If both case requirements are for a nominative, then there's no problem: the relative pronoun would then be the nominative "whoever".
If both case requirements are for an accusative, then supposedly there's no problem: the relative pronoun would then be the accusative "whomever". But that would seem to be somewhat formal in style.
If both case requirements conflict with each other (like they do in your example), then there is a problem: usually, the relative pronoun chosen would be the nominative "whoever". But many wouldn't find that fully acceptable, especially not in a formal style. (This is consistent with the general rule-of-thumb: Use "who" or "whoever" unless it sounds off to your ear.)
ASIDE: You will see some "grammar sources" (e.g. grammar usage manuals, "pop grammarians", grammar sites on the internet, retired teachers who wrote a grammar book, etc.) that, more or less, have the "rule" that the relative pronoun takes the case that is required by the relative clause. But obviously that "rule" isn't so good. For instance, "Whomever he marries will have to be very tolerant" (same as [18.iii] in CGEL) doesn't sound too smooth and is at best very questionable. And there are some sources that have a hybrid "rule", but the basis of their "rule" is that the relative clause usually decides on the pronoun's case. I'm telling you all this just so you'll know.
Back on topic.
Here's some related info from a vetted grammar source, the 2002 reference grammar by Huddleston and Pullum et al., The Cambridge Grammar of the English Language (CGEL). On pages 1073-4:
Case
The pronoun what in [17] -- [F.E.: 17. what(i) she wrote __(i)] -- is simultaneously head of the whole NP and object (in prenuclear position) in the relative clause. In constructions with personal who/whom and whoever/whomever, the pronoun has to satisfy the case requirements of both the relative clause and the matrix clause in which the whole NP is functioning. Compare:
[18]
i. [Whoever is responsible for the damage] must pay for it.
ii. He will criticise [whomever she brings home].
iii. (?) [Whomever he marries] will have to be very tolerant.
iv. (?) She lunches with [whomever is going her way after morning classes].
In [i] both the whole NP (bracketed) and the relativised element (italicized--F.E.) are subject of their respective clauses: the nominative form matches both requirements.
In [ii] both the whole NP and the relativised element are objects, and the accusative is fully acceptable though somewhat formal in style.
In [iii-iv], however, there is a clash between the function of the whole NP and that of the relativised element -- respectively subject and object in [iii], object of a preposition and subject in [iv] -- and the result is at best very questionable. Whoever would be preferable in both, but many would regard it as less than fully acceptable in formal style.
.
Since your question is about the case of the relative pronoun "who(m)ever", you'll probably also need to be aware of the "free choice" construction. Here's some related info, from CGEL page 1075:
The free choice construction (You can do whatever/what you want)
[23]
In fused relatives like these, the referent of the (overt or understood) subject of the matrix clause is given the freedom to choose: in [i] it is for you to decide who you invite, and in [ii] it is for Liz to choose where she goes. We have used 'free choice' to label one sense of any, and constructions with any + integrated relative are very close in meaning to the above: compare Invite anyone you like and Liz can go anywhere she wants (but see Ch. 5, &7.14, fn 32 for a slight difference).
There is no detectable difference in meaning between the -ever and simple forms in this construction, and since who(m), which, and how can hardly occur in other kinds of fused relative, it is plausible to see the -ever as here omissible by virtue of being redundant: the "any whatever" meaning is entailed by the free choice and does not have to be explicitly expressed in the relative word. The verb in this construction belongs to a small class consisting primarily of choose, like, please, want, wish, and is interpreted as if it had a clausal complement -- e.g. for [ii] "She can go wherever she wants to go." Note, then, that Sack who you like does not mean "Sack the pserson that you like", but "Sack whoever you care to sack". However, the distinctive syntactic property of the construction -- that of allowing certain simple forms which do not normally occur in fused relatives -- is generally restricted to the case where the clausal complement is merely implicit, so that we have Invite who you want, but hardly (?) Invite who you want to invite. Moreover, please does not in fact license clausal complements: Go where you please, but not, in Present-day English, (*) He pleased to go to Paris. And with like the meaning differs aspectually from that of a construction with an infinitival complement. I like to take the biggest portion, for example, implies repeated taking (cf. Ch. 14, &5.6.1), whereas Take which you like does not. There are also constraints on the matrix clause. For example, the fused relative must follow the matrix verb: I'll invite who you like but not (*) Who you like I'll invite. (fn 17)
.
There's more info in CGEL that's somewhat related to your topic of fused relatives, but perhaps this is enough for now.
University libraries should have a copy of CGEL, and probably also city libraries too. To get a good handle on today's standard English, both reference grammars will often come in handy: the 2002 reference grammar by Huddleston and Pullum et al., The Cambridge Grammar of the English Language (CGEL), and the older 1985 reference grammar by Quirk et al., A Comprehensive Grammar of the English Language.
Note that the "2005 textbook ASITEG" is the 2005 textbook by Huddleston and Pullum, A Student's Introduction to English Grammar.
