when phatic communication isn't (... or is it?)

HEY, ANYTIME. No biggie. Don't worry about it. Psycholinguistic research reveals that we choose our gratitude acknowledgements pragmatically, proving that such language isn't phatic (devoid of content) after all. University of Western Ontario psychologist Albert Katz suggests men, in particular, may use "anytime" to convey doiminance by signaling they have sufficient means to do the favor again in the future.  -- Conrad McCallum, Psychology Today, March 1, 2008 (emphasis added)

I'm assuming that we would refute the idea that "phatic communication is devoid of content".  Then again, when I just tried to explain what I'm doing research on to the research librarian at the BL, I immediately put it in almost the same simplistic terms.

1. "How are you doing?"

2. "Oh really badly, actually."

3. "Gosh, what's wrong?"

4. "I just stubbed my toe!"

My claim was that Locutions 1-3 are all "phatic" insofar as they contribute to establishing a channel, and Locution 4 is not, insofar as it is actually content-full.  But maybe my illustration (or analysis) is "wrong"?

Anyway, with that thought in mind, the research by Albert Katz is described in more researchy terms as follows:

Abstract. Two studies examined whether the acknowledgments given to an expressed thanks for performing a favor was merely phatic in nature, that is, does not convey information but just serves a social role in establishing and maintaining relationships. We were especially interested in a non-literal form of acknowledgment, responses such as “anytime,” which, if taken literally, invites unwanted intrusions into one's life. In the reported studies the cost of the favor was manipulated (in terms of effort, resources or time needed to perform the favor) and whether the gender of the person performing the favor was the same or opposite as that for whom the favor was done. Across the two studies, the non-literal acknowledgment was less likely to be employed for high-cost favors, was more likely to be recalled compared to other acknowledgments generated at the same base rate and was used differently by female and male participants. These data demonstrate that people moderate their thanks as a function of the cost of the favor and to whom they are speaking, results indicating that gratitude acknowledgments cannot be considered mere phatic communication. In a more general way, these data indicate the importance of considering social knowledge in nonliteral language usage.

From: Katz, Albert N., Melony Lenhardt, and Kirsten Mitchell. "On acknowledging thanks for performing a favor." Metaphor and Symbol 22.3 (2007): 233-250.