Requests are extremely common. In fact, some researchers (Tomasello, 2008) have argued that requests are one of the three core actions comprising human communication, along with offering and showing.

Some requests are “small”, such as asking your partner to pass the salt at the dinner table. Other requests are considerably larger, such as asking someone whether you can crash on their couch for multiple months. People might feel more comfortable making a small request than a big one––and recipients might also be more likely to grant those requests. Jerry Seinfeld even has a comedic bit about how the size of a request (or “favor”) seems to correlate with the size of the pause that someone takes before making the request:

There’s two types of favors, the big favor and the small favor. You can measure the size of the favor by the pause that a person takes after they ask you to ‘do me a favor.’ Small favor, small pause. Can you do me a favor, hand me that pencil? No pause at all. Big favors are, ‘Could you do me a favor…’ (huge pause, followed by closing credits.)

Importantly, requesters have more communicative resources at their disposal than simply the size of their pause. Human language is incredibly flexible, allowing us to ask for the same thing in multiple ways. How do requesters determine which form to use for their request, and how does their decision influence a recipient’s view of the requester? This is exactly the question that I took up with Dr. Federico Rossano, a professor in the Cognitive Science Department at UC San Diego and PI of the Comparative Cognition Lab; the resulting paper was just published this year in Discourse Processes (Trott & Rossano, 2020).

How do we ask for things?

Consider the number of ways that you might ask for a ride to the airport:

• “Give me a ride to the airport.”
• “Can you give me a ride to the airport?”
• “I was wondering if you could possibly give me a ride to the airport?”
• “I’m not sure how I’m getting to the airport…”

The first is what’s known as a direct request. At least in English, direct requests are formatted as imperative statements (i.e., commands).

The second option is syntactically formatted as a yes/no question about the hearer’s ability, but of course we understand this to be a request for a ride; let’s call this a modal request (because of the use of the modal verb can). The third option is similar, but further embeds this modal request in another construction (e.g., “I was wondering if…” or “do you think you could…”); let’s called these embedded modal requests.

And finally, the fourth option doesn’t explicitly ask for a ride to the airport at all––it’s merely an assertion about the state of the world. These indirect statements are one of the most puzzling forms of requests, because they require the recipient to infer what it is the requester is actually asking for.

Of course, there are a number of other ways that one might ask for a ride to the airport. But this rough taxonomy is useful for distinguishing some of the primary grammatical forms that a speaker might use to ask for something. Now we’re in a position to ask: which factors influence a requester’s choice of form?

Current work: the role of entitlement

Our paper was most directly inspired by recent research in Conversation Analysis, which suggests that a request’s formatting is influenced by the requester’s entitlement to make a request––i.e., the requester’s expectations that the recipient is willing, obligated, and/or able to fulfill their request.

There’s been extensive work in Conversation Analysis on this topic (Curl & Drew, 2008; Heinemann, 2006; see the paper for additional references), examining the relationship between entitlement and request formatting. The basic consensus is that increased entitlement licenses the use of more direct requests (e.g., “Give me a ride to the airport”), while requesters with less entitlement are more likely to embed their request in a question or elaborated statement (e.g., “Would it be possible for you to give me a ride to the airport?”).

To date, however, this relationship has not been investigated in a controlled setting, making it harder to determine whether the underlying association is truly causal. Of course, while laboratory studies have greater experimental control, they have much less ecological validity than an analysis of naturally-occurring requests. In our view, both approaches are invaluable for studying a phenomenon––analyses of actual conversations are important for establishing and studying a phenomenon in an ecologically valid setting, and controlled experiments are important for establishing causality mechanisms.

Importantly, a requester’s choice of form is a also kind of social action; thus, it’s not just that entitlement influences how requesters make their requests––a requester’s choice of form acts as a kind of signal or display of their entitlement. This makes an additional prediction, which is hard to test “in the wild” but can easily be tested in the lab: the recipients of requests might adjust their preferences for different request forms depending on the entitlement of the speaker. For example, they might view a direct request as inappropriate when the requester has low entitlement, but perhaps that same request would be perfectly licensed if the requester has relatively higher entitlement.

Additional accounts

Intuitively, there are a number of factors that could play a role in request formatting. As noted in the Introduction, one such factor might be the size of the request, i.e., the degree of imposition. In fact, politeness theory argues that the weightiness of a request can be decomposed into three main factors:
1) A request’s degree of imposition;
2) The power differential between requester and recipient;
3) The social distance between requester and recipient.

According to this account, requesters are more likely to adopt various politeness strategies for “weightier” requests. I.e., requesters should be more likely to use a polite form (like an Embedded Modal) for more imposing requests.

Our primary goal in the current work was not to adjudicate between Politeness Theory and the Entitlement Account described above; I think they are mutually compatible. “Entitlement” could conceivably manifest along each of these dimensions––one of our experimental manipulations of entitlement was, in fact, a manipulation of the power differential between the requester and recipient. In our view, however, entitlement can also manifest in additional ways, such as the urgency of the request (more urgent = more entitled), or the interactional history of the requester and recipient (i.e., if the requester has already performed a number of favors for the recipient, they might be more entitled to a favor of their own).

But because these factors might overlap, we did collect normed judgments for each of our experimental scenarios about each of those factors––that allows us to ask whether human ratings of entitlement explain formatting preferences above and beyond the factors comprising weightiness alone.

Experiment and results

Methods

We ran three online studies, which all had a similar methodology.

In Experiments 1-2, participants read a series of passages, each describing a scenario where the participant must make a request for something. For example, in one scenario, the main character was moving into an apartment with a friend, and had to ask for help carrying a very heavy desk. In Experiment 1, participants were given a free-response box to type in their request; they were given no instructions as to how this should be formatted. In Experiment 2, participants selected from four possible response options, each corresponding to the four forms described earlier (Direct, Modal, Embedded Modal, Indirect Statement).

Experiment 3 used all the same stimuli, but instead of asking participants to make a request, each passage ended with the other character making a request, which again, might be formatted as a Direct statement, Modal, Embedded Modal, or Indirect Statement.

Independent and dependent variables

Importantly, there were two versions of each passage: a High Entitlement and Low Entitlement version. “Entitlement” was operationalized along several dimensions:

• The social relationship between the characters (i.e., encompassing both “power differential” and “social distance” from the “weightiness” definition earlier);
• The urgency of the request;
• Whether one character “owed” the other for a previous favor.

Thus, our main independent variable of interest was entitlement. As noted earlier, we also conducted a norming study in which participants rated each passage according to the requester’s entitlement, power differential with the recipient, social distance with the recipient, and degree of imposition of the request.

In terms of our dependent variables, we were interested in both request formatting (for Experiments 1-2), as well as appropriateness judgments (for Experiment 3).

For Experiment 1, we asked multiple people to categorize each participant’s responses as one of the four possible request formats. Each annotator was not privy to the condition in which the response was produced. Experiments 2-3 did not require this step, as participants were asked either to select a given response option (Experiment ) or rate the appropriateness of a request on a scale from 1-5 (Experiment 3).

Results

I’ll start by comparing Experiments 1-2, since those were a very similar methodology but varied in the way participants “produced” their request.

Experiments 1-2

In both studies, our primary analysis asked whether variance in request formatting was predicted by entitlement. In Experiment 1, we found––to our surprise––that there was no significant relationship between these variables. There were some small differences in the rate of different responses chosen across conditions; e.g., 32% of the requests in the High Entitlement condition were Direct, while only 25% were Direct in the Low Entitlement condition. But importantly, this difference did not prove significant.

In contrast, we did find a strong (and very significant) relationship between entitlement and formatting in Experiment 2, 33% of requests in the High Entitlement condition were Direct, while only 13% were Direct in the Low Entitlement condition. Conversely, only 25% of requests in the High Entitlement condition were Embedded Modals (e.g., “I was wondering if you could give me a ride to the airport”), but Embedded Modals accounted for 41% of responses in the Low Entitlement condition. Unlike Experiment 1, these differences did prove to be significant.

These results seem hard to reconcile––why would the relationship be so strong in Experiment 2, but weak or nonexistent in Experiment 1? The obvious difference between these experiments is the task: participants in Experiment 1 produced their response in a free-response box, while participants in Experiment 2 selected among multiple options.

One very speculative explanation is that requesters producing a request “from scractch” (as in Experiment 1) converge on a relatively small set of “default” forms (as evidenced by a strong preference for Modal requests); any effect of entitlement might occur around the margins, and thus be harder to detect. Determining whether or not this is true would require running a larger experiment with more experiments and more statistical power.

Experiment 3

The results in Experiment 3 were much more robust than Experiment 1. Participants’ appropriateness judgments were strongly predicted by the interaction between the request form that a character used and their degree of entitlement.

This result is probably clearest with an example. The appropriateness of Direct requests (e.g., “Give me a ride to the airport”) changed substantially between Low Entitlement (M = 3.26, SD = 1.91) and High Entitlement (M = 4.23, SD = 1.77) conditions. In other words, the same request was perceived as differentially appropriate as a function of how much entitlement the speaker had to make that request.

Intriguingly, participants also displayed overall preferences towards particular request forms, as measured by appropriateness judgments. They preferred Embedded Modals and Modals overall. In fact, the Embedded Modal form (“I wonder if you could give me a ride to the airport”) was the only form whose appropriateness was not affected by entitlement.

In other words, recipients generally prefer these more elaborate request forms regardless of the situation––but shorter, more direct forms are licensed when the speaker has greater entitlement.

The Takeaway

So what did we learn from this?

First, Experiment 2 serves as a proof of concept that participants do recognize normative mappings between entitlement and request formatting. This is consistent with the predictions of past work. More puzzling, however, is the finding that entitlement was not a significant predictor of formatting in Experiment 1. I mentioned one very speculative explanation earlier on; future work might benefit from rerunning Experiment 1 with a larger sample to determine whether this is an issue of statistical power or whether there really is no effect using that methodology. And if the latter is true, it raises the question of under which conditions requesters actually act on these normative mappings.

Experiment 3 is, as far as I know, the first experiment that asks whether entitlement affects a recipient’s preferences for various request forms. We found quite convincing evidence in the affirmative: specifically, more direct forms are licensed when the requester was higher entitlement to make a request.

Of course, it’s worth noting that none of these experiments are particularly naturalistic. A legitimate criticism of any vignette study is that its lack of ecological validity makes it hard to generalize to more “real-world” scenarios. I’m quite sympathetic to this criticism, but as I noted earlier, I think these more tightly controlled scenarios are useful complements to more naturalistic studies. No study should be seen as the “final word” on a given scientific hypothesis. But taken together, a set of findings start to provide more or less evidence for that hypothesis.

In this case, it appears that requesters and request recipients alike can recognize systematic mappings between entitlement and request formatting––the question that remains is whether and under what conditions requesters regularly act on these norms “in the wild”. Past work (Curl & Drew, 2008; Heinemann, 2006; and much more!) suggests that they do; Experiment 1 suggests they do not do so in a controlled setting, or only do sometimes. One fruitful direction (in addition to replicating Experiment 1 with a larger sample) would be to conduct a similar study in-person (once it’s safe to do so), which might come closer to replicating the conditions under which people produce requests “in the wild”.

The other obvious area for future work––as it so often is––is individual differences. Across all experiments, some participants showed a stronger effect of entitlement than others. And some participants failed to adjust their preferences at all. What accounts for these differences? Individuals (and cultures) might vary in several relevant ways: 1) different “default” distributions over the possible forms; 2) different conceptualizations of what makes someone more or less entitled; and 3) diferent mappings from entitlement to request forms. These individual differences might also intersect with various forms of social identity, such as gender.

References

Curl, T. S., & Drew, P. (2008). Contingency and action: A comparison of two forms of requesting. Research on language and social interaction, 41(2), 129-153.

Heinemann, T. (2006). ‘Will you or can’t you?’: Displaying entitlement in interrogative requests. Journal of Pragmatics, 38(7), 1081-1104.

Trott, S., & Rossano, F. (2020). The Role of Entitlement in Formatting Preferences Across Requesters and Recipients. Discourse Processes, 1-22.