Communicating Imperatives Requires Psychological Closeness But Creates Psychological Distance

Abstract

This article evaluates the psychological correlates of imperative speech through pronouns. We demonstrate that people communicate with more collective immediacy (“we” words) when using imperatives than nonimperatives in an experiment (Study 1, N = 828) and field studies of American politicians (Study 2a: N = 123,678 speeches), and Joseph Stalin (Study 2b: N = 593 speeches). However, respondents experience a psychological distancing effect after an imperative (fewer “I” words). This experimental pattern (Study 3: N = 852) also holds in the field using U.S. Supreme Court dissents from the Roberts Court (Study 4: N = 644). Exploratory findings suggest that third-person plural pronouns (“they” words) are used more when communicating imperative speech relative to nonimperative speech. Our evidence supports an interpersonal imperatives asymmetry: imperatives demand psychological support when communicating how the world must be, but they undermine the autonomy of respondents. Social and psychological implications of these data are discussed.

Keywords

imperatives psychological distance automated text analysis big text data

Language patterns offer a lens into social and psychological processes that can provide rich data about people. For example, word patterns reveal authorship signatures (Boyd, 2018), changes to political speech and culture (Jordan et al., 2019), and how emotions are expressed over time (Golder & Macy, 2011). From these studies, we can gather psychological information about what communicators are thinking and feeling. One advantage of evaluating language is that word patterns are more recorded now than any period of history and they can be evaluated with many approaches to document human behavior (Salganik, 2017).

Much of the research evaluating text-as-data from a social science perspective attempts to identify how language patterns correlate with psychological processes, rather than cause them. For example, researchers use language to evaluate how people lie and tell the truth (Markowitz & Hancock, 2014, 2016), with verbal cues such as negative affect reflecting dishonesty versus truth-telling (Hauch et al., 2015). In another example, researchers often evaluate how words indicate psychological well-being (e.g., depression). People who are depressive tend to focus inward, on the self, and have a negative disposition relative to those who are nondepressive (Weintraub, 1981). Such characteristics also appear in language (Rude et al., 2004; Tackman et al., 2019), suggesting that words can reveal the internal processes of what people are thinking, feeling, and experiencing psychologically.

The current work draws on this approach to investigate how encoding and decoding imperative speech is associated with psychological information indicated by function words. We take a bidirectional view of this phenomenon by first evaluating how people communicate imperatives and determining the psychological requirements of imperative speech through pronouns. Second, we assess how imperatives can affect a respondent. Our evidence is some of the first to indicate that pronouns support the communication process of encoding imperatives (Studies 1-2b), but they mark an undermined self-focus when responding to and decoding imperatives (Studies 3-4).

This work is timely and important for several reasons. Imperatives comprise a nontrivial amount of daily speech, but their value is often appraised in terms of functionality. Particularly, research has evaluated how and why people communicate imperatives (Austin, 1975; Clark, 1996), with less work focusing on their psychological requirements during the encoding process and their impact on respondents during the decoding process. Our work examines this bidirectional effect directly through the lens of pronouns. Since imperatives can direct or sanction behavior, we are interested in the verbal indicators that make imperative speech possible for communicators (encoders), and those that indicate imperative speech might be psychologically problematic for respondents (decoders). If we only consider how imperatives are communicated and ignore their impact on others, we fail to investigate such communication processes holistically. Finally, this work is practically important because people often take direction from others, such as politicians, bosses, or elders. It is presently unclear how those who communicate imperatives might support their views with other verbal markers, nor is it well understood how imperatives might undermine decoders by restricting their agency.

The Psychology of Language: Content and Style

The idea that imperatives are associated with psychological information that reflects encoding and decoding processes is supported by other work using words to indicate how people are thinking and feeling. Word patterns correlate with intelligence (Pennebaker et al., 2014), social status (Kacewicz et al., 2014; Markowitz, 2018), and personality traits (Ireland & Mehl, 2014). In general, word patterns can be categorized into two classes (Boyd & Pennebaker, 2015): content words (e.g., nouns, adjectives) and function words (e.g., articles, prepositions, pronouns). Content words form the topics and ideas that people remember most in a sentence (e.g., “The boy who cried wolf”), which can help identify themes in a collection of texts (Blackburn et al., 2018; Markowitz & Griffin, 2020). Function words, on the other hand, are also called style words (e.g., articles, prepositions, negations, pronouns) and bridge communication content. There are less than 500 function words in English (Chung & Pennebaker, 2007), but they are pervasive in everyday speech. Research suggests function words indicate a variety of psychological processes ranging from social status to deception (Pennebaker, 2011).

Examples of function words include articles (e.g., a, the), which are markers of concreteness (Markowitz & Hancock, 2016) and correlate positively with persuasion outcomes such as the probability of receiving an online loan from the crowd (Larrimore et al., 2011). Pronouns are common function words as well, often revealing where people focus their attention. Both experimental and observational evidence suggests high-status people tend to use more collective references (e.g., we, our, us) than low-status people since their focus needs to consider more people than the self (Kacewicz et al., 2014). Finally, contrary to popular belief, narcissism is not indicated by an increased self-focus through “I” pronouns, but instead, an increase in second-person pronouns such as you or your (Holtzman et al., 2019). The prior evidence and general theory behind function words offers that they reflect how people are communicating, rather what they are communicating about. Our study adopts this perspective to test the connection between imperative speech and pronoun use during the encoding (communicating) and decoding (responding to) processes.

Psychological Properties of Imperatives

Imperatives are often considered commands, which get an audience to perform an action via content words (e.g., “pass the salt,” “go to your room”). Less is known about how imperatives are represented through function word patterns, however, which can reveal aspects of our psychology during the process of communicating and responding to such information. There is some evidence that suggests imperative speech relates to psychological characteristics such as personality. In one investigation, Mairesse et al. (2007) observed that commands were negatively associated with conscientiousness that was assessed from an observer’s perspective, but positively associated with emotional stability as evaluated from self-report data.

Perhaps one reason for the scarce empirical connection between function words and imperative speech is that the imperatives literature originates from psycholinguistics (Moldovan et al., 2011) and speech act theory (Austin, 1975). In general, speech act theory considers how people accomplish tasks with words. Most evaluations of imperatives from this tradition assess their functional value and what people hope to accomplish by using them. As Clark (1996) identifies, imperatives “get the listeners to do things” and “commit the speaker to doing things” (p. 134). Speech acts can also be declarative by “giving an authoritative decision about some fact” (Moldovan et al., 2011, p. 24). This does not imply that an action immediately follows an imperative statement, but rather, a future action might be sanctioned or requested through an imperative statement compared with a nonimperative statement. Therefore, imperatives are social, purposeful, and goal-oriented (Salaita, 2005), helping to direct action or firmly assert a speaker’s stance on an issue or behavior (D’Andrade & Wish, 1985).

While much has been written about why imperatives and speech acts are functionally important for interpersonal tasks (e.g., encouraging another person to do something; Austin, 1975; Clark, 1996; Moldovan et al., 2011), it is unclear how imperatives connect to communication style and relate to other language dimensions that might reveal their psychological value. Discovering the connection between imperatives and function words—specifically pronouns—is crucial in our research. We seek to extend speech act theory by understanding the psychological correlates of communicating and responding to imperatives.

Several function words in English have imperative characteristics, including must, need, and should (also called modal auxiliary verbs), where the speaker suggests or authorizes behavior. Examples of imperatives demonstrate their social and purposeful nature. In response to the possibility that electronic voting has affected how we elect officials in the United States, President Jimmy Carter (2017) suggests, “We must accommodate these changing times while holding true to our unchanging principles—equality, justice and freedom for all.” Carter’s imperative (e.g., must) communicates a call-to-action that is firm and difficult to counterargue (Bénabou et al., 2018). That is, few people would object to the proposition that equality, justice, and freedom are important American values. In another example, President Barack Obama (2007) states, “We shouldn’t be providing billions in taxpayer-funded giveaways to private banks; we should be providing an affordable, accessible college education to every American.” This stance is also difficult to counterargue by including imperative indicators (e.g., shouldn’t, should), since most (young) Americans agree that access to college education is a positive social reform (Bauer-Wolf, 2019). The Carter (2017) and Obama (2007) examples show how imperatives can verbally direct action and are also associated with markers of social inclusion and immediacy (e.g., pronouns).

In addition to the functional value of imperatives established by speech act theory (Austin, 1975; Clark, 1996), we argue that imperatives are also associated with psychological characteristics that aid in their mission to sanction or request an action. We begin by suggesting that pronouns can provide a speaker with the support needed to appear credible and a part of the request. However, when imperatives are received, responded to, and decoded, they can create a psychological distancing effect.

Imperatives as Credibility Intensifiers

Mairesse et al. (2007) estimated that commands and imperatives represent nearly 4% of natural speech acts. Imperatives are therefore a nontrivial form of communication that can reflect the psychology of the speaker and change how a respondent reacts psychologically as well.

Imperatives are relationship-dependent and effective to the degree that the respondent deems the speaker trustworthy or credible (Bénabou et al., 2018). An imperative is taken seriously if it originates from a credible source and one way to assess credibility through language is to evaluate personal pronouns (Carbone, 1975). In academic papers, for example, first-person pronouns aid in amplifying an author’s credibility. Hyland (2003) suggests that first-person plural pronouns give the writer “the right to speak with authority, emphasizing the importance that should be given to his or her unique procedural choices or views” (p. 257). Therefore, sanctioning, requiring, or requesting another person to do something can be supported psychologically by amplifying the rate of first-person plurals, since these pronouns encode the speaker and addressee in the same action or statement (De Fina, 1995).

Relatedly, to get another person to do something, one strategy might be to decrease the psychological distance (increase the immediacy) between the speaker, audience, and imperative (Wiener & Mehrabian, 1968). This can be accomplished by amplifying the rate of first-person plural pronouns as well (Liehr et al., 2004). For example, suggesting that “we must,” “we should,” or “we want” to accomplish some task draws the speaker closer to the audience (Bull & Fetzer, 2006; Proctor & Su, 2011). More recent evidence suggests that first-person pronouns are also effective self-presentation markers. For example, successful online political posts (e.g., those that receive a high number of engagements) tend to contain high rates of first-person pronouns compared with less successful posts (Nave et al., 2018). The authors argue that pronouns signal positive self-presentation characteristics and can lead to positive downstream effects (e.g., convincing someone to do something, such as liking a post). Taken together, we expect a positive relationship between imperatives and first-person plurals to reflect the idea that people can amplify their credibility, boost self-presentation, and decrease their psychological distance to the audience in the process of sanctioning or requesting an action.

Hypothesis 1: Communicating an imperative will be positively associated with the rate of first-person plural pronouns (“we” pronouns).

Imperatives as “Cognitive Straightjackets.”

Imperatives leave little room for contingencies (Bénabou et al., 2018), since they often describe how the world must or should be. This position can constrain the autonomy of a respondent and limit their willingness to express an opinion after an imperative (Mazarr, 2016). We also expect this psychological experience to be reflected in word patterns such as pronouns.

As Hart and Childers (2004) suggest, pronouns index where the speaker views himself or herself in an action. Therefore, after responding to an imperative, we expect a reduction in self-focus, because a person’s ability to feel like they can act is constrained. In this way, imperatives can be perceived as “cognitive straightjackets” (Bénabou et al., 2018, p. 30).¹ The negative relationship between imperatives and self-references for decoders suggests a psychological distancing effect revealed through function words (Cohn et al., 2004). Since imperatives can constrain one’s autonomy (e.g., a person’s belief that they have a say in the world or can act independently), we expect more psychologically distant responses from communicators in the imperative decoding process.

Hypothesis 2: Responding to an imperative will be negatively associated with the rate of first-person singular pronouns (“I” pronouns).

While there are more pronoun categories than first-person plural and first-person singular, our hypotheses consider these two language dimensions because prior work has documented their substantial connection to psychological processes (Boyd & Pennebaker, 2015; Chung & Pennebaker, 2007; Pennebaker, 2011). In the spirit of research transparency and to remain consistent with other evaluations (Kacewicz et al., 2014), we also report how communicating and responding to imperatives affect four other pronoun categories in exploratory analyses. All findings are reported in the main text and Table 1 across six total pronoun categories: first-person singular (henceforth called “I” pronouns), first-person plural (henceforth called “we” pronouns), second-person pronouns (henceforth called “you” pronouns), third-person singular (henceforth called “she/he” pronouns), third-person plural (henceforth called “they” pronouns), and impersonal pronouns. We renamed the pronoun categories in the remainder of the article out of readability concerns.

Table 1.

A Priori and Exploratory Results Across Studies.

Study 1	Pronoun	Example	Imperative prompt		Nonimperative prompt		t	p	d
Study 1	Pronoun	Example	M(%)	SD(%)	M(%)	SD(%)	t	p	d
	First-person singular	I, me, my	0.38	1.24	0.40	1.24	−0.24	.814	0.02
	First-person plural	we, our, us	5.73	5.02	4.90	4.70	2.44	.015	0.17
	Second-person	you, your	0.55	1.74	0.58	2.00	−0.18	.861	0.02
	Third-person singular	he, him, she	0.05	0.43	0.05	0.29	0.27	.788	<0.01
	Third-person plural	their, they	1.11	1.93	0.84	1.65	2.17	.031	0.15
	Impersonal pronouns	it, that, who	6.36	3.71	6.52	3.99	−0.60	.547	0.04
Study 2a	Pronoun	Example	β	SE	t	p	R²c
	First-person singular	I, me, my	0.28	0.003	106.02	<.001	0.31
	First-person plural	we, our, us	0.37	0.002	149.80	<.001	0.36
	Second-person	you, your	0.24	0.002	97.10	<.001	0.42
	Third-person singular	he, him, she	−0.17	0.003	−62.67	<.001	0.25
	Third-person plural	their, they	0.28	0.003	103.35	<.001	0.22
	Impersonal pronouns	it, that, who	0.38	0.002	151.78	<.001	0.36
Study 2b	Pronoun	Example	β	SE	t	p	R²adj.
	First-person singular	I, me, my	−0.04	0.047	−0.88	.378	0.28
	First-person plural	we, our, us	0.18	0.050	3.63	<.001	0.16
	Second-person	you, your	−0.07	0.050	−1.37	.172	0.16
	Third-person singular	he, him, she	0.09	0.049	1.90	.057	0.20
	Third-person plural	their, they	0.16	0.053	2.94	.003	0.06
	Impersonal pronouns	it, that, who	0.35	0.036	9.79	<.001	0.58
Study 3	Pronoun	Example	Imperative prompt		Nonimperative prompt		t	p	d
Study 3	Pronoun	Example	M(%)	SD(%)	M(%)	SD(%)	t	p	d
	First-person singular	I, me, my	3.53	2.61	3.97	2.82	−2.38	.018	0.16
	First-person plural	we, our, us	0.19	0.92	0.10	0.64	1.62	.106	0.11
	Second-person	you, your	0.72	1.86	0.61	1.80	0.83	.406	0.06
	Third-person singular	he, him, she	0.02	0.36	0.13	0.87	−2.44	.015	0.17
	Third-person plural	their, they	1.93	2.83	1.66	2.58	1.48	.138	0.10
	Impersonal pronouns	it, that, who	7.45	4.34	7.12	4.47	1.10	.272	0.07
Study 4	Pronoun	Example	β	SE	t	p	R²c
	First-person singular	I, me, my	−0.09	0.039	−2.24	.026	0.14
	First-person plural	we, our, us	0.08	0.037	2.20	.028	0.30
	Second-person	you, your	−0.10	0.038	−2.69	.007	0.26
	Third-person singular	he, him, she	−0.04	0.038	−1.08	.283	0.36
	Third-person plural	their, they	0.03	0.038	0.85	.394	0.30
	Impersonal pronouns^a	it, that, who	0.18	0.039	4.59	<.001	0.05

Note. SE = standard error. Studies 2a, 2b, and 4 relate each pronoun category to the imperative dictionary in separate regression models. The field study data were natural log-transformed out of skewness concerns.

The author and case random effects were removed from the impersonal pronouns model in Study 4 to aid in model convergence.

We also acknowledge that imperatives come in many forms that are not indicated through function words (e.g., directives or commands have rich content words such as the case of “pass the salt”; Clark, 1996). We use pronouns to evaluate how imperatives associate with style markers, a contribution to speech act theory and research in the psychology of language tradition, since the majority of imperative research has focused on content.

Study 1

Method

Power

For each experiment, we determined sample sizes based on an a priori power analysis, which required a minimum of 788 participants to detect a small effect (Cohen’s d = 0.20) powered at 80% (α = .05; two-tailed).

Participants with short texts (≤15 words) were excluded to prevent low word counts from overstating the prevalence of word frequencies. The experiments received institutional review board approval from the authors’ institution and reproducibility files are available on the Open Science Framework (https://osf.io/w679k/). Note, raw LIWC outputs are provided on the Open Science Framework and field study data were natural log-transformed out of skewness concerns.

Participants

Participants in Study 1 (N = 828) were recruited from TurkPrime, an extension of Amazon Mechanical Turk, and paid $0.25 for their time in a short experiment. Most participants were female (n = 537, Male n = 284, Other n = 5, some participants did not respond to demographic questions) and gender was evenly represented across conditions, χ²(2) = 2.43, p = .297. On average, participants were 41.94 years old (SD = 13.46) and evenly represented across our experimental conditions as well, t(818) = 1.06, p = .291. These data suggest that random assignment was indeed successful.

Procedure

After consenting to participate via the Qualtrics survey interface, participants clicked a reCAPTCHA button to reduce the possibility of receiving nonhuman responses. Then, participants were randomly assigned to one of two experimental conditions: (a) imperative prompt or (b) nonimperative prompt. Those assigned to the imperative prompt were provided with the following instructions:

Imagine you are giving a speech to college students. Your speech topic is: Why America must protect itself from acts of terrorism. Write the opening paragraph of this speech, defending why you believe that America must protect itself from terrorism.

Those assigned to the nonimperative prompt were provided with the following instructions:

Imagine you are giving a speech to college students. Your speech topic is: Why America benefits from protecting itself from acts of terrorism. Write the opening paragraph of this speech, describing why you believe that America can benefit from protecting itself from terrorism.

Therefore, participants were put into an imperative or nonimperative mind-set and told to write. The writing task was untimed, after which participants provided demographic data and then exited the survey.

Automated Text Analyses

We analyzed the participant’s writing style with Linguistic Inquiry and Word Count (LIWC: Pennebaker et al., 2015), an automated text analysis tool that quantifies language patterns. LIWC contains a dictionary of words that reflect a range of social (e.g., words related to friends, family), psychological (e.g., words related to cognitive processes, emotion), and part of speech categories (e.g., function words). This tool calculates word frequencies as a percent of the total word count. For example, the phrase “We the people of the United States” contains seven words and increments several LIWC categories, including but not limited to: first-person plural pronouns or “we” pronouns (we; 14.29% of the total word count) and articles (the; 28.57% of the total word count). We used LIWC in all studies, and the language dimensions were drawn from the standard LIWC2015 dictionary unless otherwise noted. There were no word count differences across conditions, t(768.77) = −0.39, p = .697.

“We” Pronouns

We evaluated how an imperative prompt, relative to a nonimperative prompt, affects a speaker through rates of “we” words (M = 5.34%, SD = 4.89; Q1 = 0.00%, Mdn = 4.55%, Q3 = 8.64%).

All effects across studies are summarized in Table 1.

Results

The data suggest that people communicate with more “we” pronouns after an imperative prompt (M = 5.73%, SD = 5.02%) compared with a nonimperative prompt (M = 4.90%, SD = 4.70%), t(826) = 2.44, p = .015, d = 0.17. Therefore, the data support Hypothesis 1: communicating and encoding an imperative requires psychological closeness in the form of collective references.

Exploratory Analyses

The exploratory data in the Table 1 suggests that the rate of “they” words was significantly higher for participants in the imperative condition (M = 1.11%, SD = 1.93%) than those in the nonimperative condition (M = 0.84%, SD = 1.65%), Welch’s t(825.89) = 2.17, p = .031, d = 0.15. Differences across conditions for the other pronoun variables did not reach statistical significance (ts < 0.60, ps > .547).

Discussion

One possible explanation for the significant “we” words finding and nonsignificant findings for many other pronouns might be that parts of the stimuli (e.g., commenting on America) facilitated a collective focus. Prior evidence suggests that people who are primed with interdependent and collective stories tend to self-describe in the same interdependent and collective manner (Gardner et al., 1999; Markus & Kitayama, 1991; Trafimow et al., 1991). The “they” pronoun effect, which mirrored the “we” pronoun effect for imperative speech, might also be an indication of social comparison between Americans and terrorists. Prior work suggests that people use high rates of “they” pronouns to “emphasize interpersonal comparison” especially among a strong collective (Van Swol & Carlson, 2015, p. 514). In this case, where Americans are describing why America must protect itself from acts of terrorism, it is perhaps reasonable to observe high rates of collective focus and “they” pronouns to reflect interpersonal comparison (e.g., describing how “we” as Americans might be against “them” as terrorists). This also might be an indication of intergroup bias revealed through language (Gustafsson Sendén et al., 2014; Perdue et al., 1990), though the idea warrants future study.

To test Hypothesis 1 outside of an experimental setting and further investigate the exploratory relationships, we performed a field study by collecting speeches from American politicians. We performed the same analytic procedure by assessing the relationship between imperatives and “we” pronouns, while also exploring how other pronoun patterns (e.g., “they” pronouns) might correlate with imperatives as well.

Study 2a

Method

Data Collection

We collected a database of text with a presumed reasonable baseline level of imperatives. We used speeches from high-ranking American politicians because leaders often direct action compared with followers (Johnson, 1994) and politicians are some of the most public leaders who can reach masses in the United States.

Our sample included public statements made by politicians (presidents and nonpresidents) indexed by the American Presidency Project (APP: Woolley & Peters, 2019). The APP is a database of oral and written statements since George Washington’s inaugural address in 1789. Each speech contains a host of metadata curated by the APP, such as the speech date, the speech type (e.g., messages, campaign documents, written statements), and the speaker (e.g., presidents or members of an administration), which served as controls in our analyses. Note, some non-presidents were indexed by the administration they served (e.g., a press briefing by Gene Sperling was labeled as William H. Clinton because he served as the Director of the National Economic Council for the Clinton Administration).

We excluded texts with less than 25 words to prevent low word counts from exaggerating the prevalence of word frequencies. Our final sample from the APP contained 123,678 speeches (103,570,195 total words; M = 837.42 words per text, SD = 1,124.88 words; Q1 = 168 words, Mdn = 361 words, Q3 = 938 words). Our data contains texts across 16 speech types and 116 speakers (e.g., presidents, vice presidents, chiefs of staff, among others).

Imperative Dictionary

Most research from speech act theory relies on humans to interpret the imperative value of a text. With over 100,000 documents, it would be unwieldy for coders to be accurate and consistent in making imperative judgments across such a large number of files (see Tausczik & Pennebaker, 2010). Therefore, we took an automated approach and created a dictionary of function words that indicate imperative speech in naturally occurring texts (e.g., words that sanction or request an action from the speaker to the audience). This is a common method, which allows complex psychological processes to be evaluated with word patterns at scale (e.g., Graham et al., 2009).

Our imperative dictionary was adapted from existing word collections. The LIWC2015 discrepancy category contains words that indicate “how the world is and how it could, should, or ought to be” (Pennebaker, 2011, p. 165). Many words in this category are modal auxiliary verbs and directives, such as must and need, but there are others that do not convey imperative thinking (e.g., if, mistake, odd). We therefore created an imperative dictionary of nine function words that attempt to indicate an author trying to “get the addressee to do something” (Clark, 1996, p. 135): must, need, needed, needing, ought, should, shouldn’t, want, and wanted.

We computed a reliability measure for our imperative dictionary using State of the Union speeches from the APP database (N = 238 speeches; 1,985,487 words).² Based on the process outlined by Pennebaker et al. (2015), each word from the imperative dictionary received a frequency score across the speeches, which was then used for a reliability calculation: intraclass correlation using standardized items = 0.41; F(237, 1896), p < .001. The index reached a reliability level consistent with other LIWC dimensions (Pennebaker et al., 2015) and was highly correlated with the discrepancy category (r = 0.88, p < .001), which allowed us to proceed with the evaluation of our texts for imperative speech. Note, while internal consistency measures for words are often lower than traditional scale measures, they remain highly significant and valid for language assessments (Tausczik & Pennebaker, 2010).

Results

The data were analyzed with linear mixed models using the lme4 package (version 1.1-21) in R (version 3.5.1), controlling for the speech year and speaker as random effects, since multiple speeches from one speaker are not independent. Speech type was entered as a fixed effect to aid in model convergence. Recall, independent and dependent variables were natural log-transformed to reduce data skewness by first adding a constant to the raw score (constant = 1) and then standardizing the value.

Imperatives positively predicted the rate of “we” pronouns (β = 0.37, SE = 0.002, t = 149.80, p < .001, R²c = 0.36. Therefore, consistent with Hypothesis 1, imperatives require psychological support that can be achieved through collective references.

Exploratory Analyses

Our exploratory findings in Table 1 suggest that imperative speech was positively associated with four pronoun categories (e.g., “I” pronouns, “you” pronouns, “they” pronouns, and impersonal pronouns) in addition to collective references, βs > .24, ps < .001, R²c > 0.22. Imperative speech was negatively associated with “she/he” pronouns, however, β = −0.17, SE = 0.003, t = −62.67, p < .001, R²c = 0.25.

Discussion

Given the scale of these data, it is perhaps unsurprising that the pronoun categories were significantly related to imperative speech. These effects reinforce the general theory and evidence from Study 1 that suggests imperative speech, at least communicated by high-ranking American politicians, is supported by an increase in immediacy markers. One possible explanation for these findings is social inclusivity—by considering all people in a communication act—which might be a social and psychological goal when politicians communicate imperatives because they are trying to direct the public into action. Evidence suggests that an event is more palatable and perceived more favorably when people have a shared, simultaneous experience (Boothby et al., 2014). Therefore, speakers may believe that an imperative is likely to be adopted if a collective, including the imperative’s author, supports the statement. The “she/he” pronoun effect also reveals that American politicians are less likely to identify specific individuals (e.g., him, her, herself) when speaking in an imperative manner, presumably because individuation can be perceived as identity threatening (for a review, see Petriglieri, 2011).

One critique of this initial observational study is that the data set only reflects politicians from the United States. To evaluate the robustness of this phenomenon and how the effects hold across nations, we collected a large sample of speeches from Joseph Stalin, former General Secretary of the Communist Party of the Soviet Union. We investigate the relationship between imperative speech and “we” pronouns (Hypothesis 1) to measure if psychological support for an imperative stance, even by an authoritarian figure, is amplified by a collective orientation. Consistent with our prior evaluations, we also explore how other pronoun patterns relate to imperative speech.

Study 2b

Method

All transcripts from Joseph Stalin were indexed by the Marxists Internet Archive and translated to English by the archivists. We collected 593 texts from 1901 to 1952, totaling 587,569 words (M = 990.84 words per text, SD = 2,559.49 words; Q1 = 91 words, Mdn = 257 words, Q3 = 762 words). The data were analyzed with LIWC, natural log-transformed to reduce skewness using the process from Study 2a, and then standardized. Consistent with our other field studies, texts with fewer than 25 words were excluded. We controlled for year as a fixed effect to account for speaking style variability over time (see Pennebaker & Stone, 2003).

Results

Consistent with Study 2a, the rate of imperatives and “we” words were positively associated, β = 0.18, SE = 0.05, t = 3.63, p < .001, R²adj. = 0.16. Therefore, the link where imperatives are supported by collective pronouns extends beyond U.S. politicians and reflects a robust social and psychological phenomenon.

Exploratory Analyses

Two statistically significant patterns emerged that were consistent with Study 2a: imperative speech was positively associated with “they” pronouns, β = 0.16, SE = 0.05, t = 2.94, p = .003, R²adj. = 0.06, and impersonal pronouns, β = 0.35, SE = 0.04, t = 9.79, p < .001, R²adj. = 0.58. The relationship between imperative speech and “she/he” pronouns was marginally significant, β = 0.09, SE = 0.05, t = 1.90, p = .057, R²adj. = 0.20, and the relationship between “I” pronouns and imperative speech was not statistically significant, β = −0.04, SE = 0.05, t = −0.88, p = .378, R²adj. = 0.28.

Discussion

Across the first three data sets in this article—one experiment and two field studies—we found that imperatives are most consistently associated with “we” pronouns and “they” pronouns. One possible explanation for these effects, also offered in Study 1, suggests that imperatives might facilitate perceptions of intergroup bias that affect and relate to the speaking style of communicators. That is, people who describe how the world must, should, and ought to be may also try to create distance between the collective and those viewed as “others.” This evidence is related to other work that suggests on xenophobic online forums, people who post more often over time and feel like they are a part of the group also tend to use more “we” pronouns and “they” pronouns (e.g., establishing the collective while distancing the out-group; Bäck et al., 2018).

Perhaps the self-references (“I” pronouns) finding from Study 2a did not extend to the Stalin case because leaders (e.g., those of high-organizational rank) tend to use more collective references than self-references as a reflection of their place in a social hierarchy (Kacewicz et al., 2014). As noted, those who are primed to think in a collective manner tend to self-describe through “we” pronouns as well (Gardner et al., 1999; Trafimow et al., 1991). To visually inspect the strength of the relationship between imperative speech and “we” pronouns across speakers from Studies 2a and 2b, please see the left panel of Figure 1. The left panel only reflects U.S. presidents from Study 2a out of space limitations.

Figure 1.

Correlations between imperative speech and pronouns.

The initial studies in this article suggest that encoding and communicating imperatives are associated with a psychological focus on the collective. The second major part of our investigation tests how respondents decode imperatives as indicated by pronouns. Next, we perform an experiment to evaluate Hypothesis 2, which proposes that an imperative (compared with a nonimperative) can create a psychological distancing effect for the respondent, specifically reflected by a reduced rate of “I” pronouns. We also continue our exploratory analyses by investigating other pronouns that correlate with the psychological experience of responding to imperatives.

Study 3

Method

Power and Participants

Participants were recruited from Amazon Mechanical Turk and paid $0.25 for their time. We used reCAPTCHA technology in the online survey to curb nonhuman responses. After excluding participants with duplicate Mechanical Turk IDs, IP addresses, and those who wrote short text (≤15 words), the final sample contained 852 participants across two experimental conditions. This number was consistent with our required sample size to detect a small effect powered at 80% (see Study 1).

Most participants were female (n = 505, Male n = 340, Other n = 6, some participants did not respond to demographic questions) and gender was evenly represented across conditions, χ²(2) = 1.71, p = .425. On average, participants were 36.45 years old (SD = 12.85) and evenly represented across our experimental conditions as well, t(842) = 0.17, p = .862. We also asked participants for a binary response if they ever applied to graduate school in the past, which was relevant to our experimental manipulation (see below). This characteristic was also evenly represented across conditions, χ²(1) = 1.07, p = .302. These data suggest that random assignment was indeed successful across our two conditions.

Procedure

After we obtained consent via the Qualtrics survey interface, we told participants that “the University of Oregon has changed its policy on graduate education to improve academic standards. In prior years, the university accepted students with average Graduate Record Examination (GRE) scores.” Following this description, participants were randomly assigned to one of two conditions, which described the University’s policy on graduate school admissions. Participants randomly assigned to the imperative condition saw the following text: “With the new policy, the admissions office has directed that students must have exceptional scores in order to be competitive for admission.” Participants randomly assigned to the nonimperative condition saw: “With the new policy, students are encouraged to have exceptional scores in order to be competitive for admission.” All participants were then instructed to “Tell us in your own words, why you agree or disagree with this policy,” and an essay text box allowed participants to type their response.

Following the untimed writing task, participants provided demographic data and were debriefed by revealing that the University of Oregon was not reconsidering its policy on graduate admissions.

Automated Text Analysis

“I” Pronouns

The rate of “I” pronouns reflects attention on the self (M = 3.76%, SD = 2.73%; Q1 = 2.03%, Mdn = 3.45%, Q3 = 5.13%). Text with a high rate of self-references suggests a more immediate, self-focus than text with a low rate of self-references.

Word count was not significantly different across conditions, t(850) = 1.42, p = .16.

Results

Supporting Hypothesis 2, participants who responded to the imperative prompt used fewer “I” pronouns (M = 3.53%, SD = 2.61%) compared with participants who responded to the nonimperative prompt (M = 3.97%, SD = 2.82%), t(850) = −2.38, p = .018, Cohen’s d = 0.16. Responding to an imperative causes a psychological distancing effect relative to responding to a nonimperative as reflected through rates of self-references. Without an imperative, people assert the self in their disclosures and presumably feel increased agency.

Exploratory Analyses

Only “she/he” pronouns revealed a significant difference between the imperative (M = 0.02%, SD = 0.36%) and nonimperative conditions (M = 0.13%, SD = 0.87%), Welch’s t(597.23) = −2.44, p = .015, Cohen’s d = 0.17. These data should be interpreted with caution, however, because baseline levels of “she/he” pronouns are low in this study. In general, responding to an imperative affects self- and singular-focus, but not collective focus.

Discussion

The effects in Study 3 revealed that decoding an imperative causes a reduction in self-focus and singular focus in general. That is, people tend to use fewer “I’ and “she/he” pronouns when responding to an imperative prompt compared with a nonimperative prompt. This reduction in pronoun singularity reflects where people are focusing their attention after an imperative compared with a nonimperative (see Pennebaker, 2011). People tend to focus less on the self or individual actors when decoding an imperative, which is a powerful reduction in immediacy and agency, plus an extension of the cognitive straightjacket theory proposed by Bénabou et al. (2018). Therefore, imperatives facilitate or require collective focus (Studies 1, 2a, and 2b), but constrain singular focus (Study 3) compared with nonimperatives.

Next, to test if imperative responses are associated with a psychological distancing and singularity-reduction effect in the field, we collected dissenting opinions from the U.S. Supreme Court (2005-2019). Dissents are written by justices who oppose the majority opinion. Therefore, they represent a prime data source to evaluate if high rates of imperatives (e.g., describing how the world must be as a result of court decisions) associate with more psychological distancing in the text (e.g., a reduced rate of “I” pronouns).

Study 4

Method

We collected Supreme Court transcripts from Justia, one of the largest online archives of legal texts. Roberts Court dissents (2005-2019) were gathered until early June 2019. We excluded mixed dissents (e.g., dissenting in part) and texts from three justices who wrote less than 10 dissents within the data collection timeframe (Justice Rehnquist, Justice O’Connor, Justice Kavanaugh). Our sample included 644 dissents across 12 justices (2,513,543 total words, M = 3,903.02 words, SD = 3,235.40 words; Q1 = 1,531 words, Mdn = 3,260 words, Q3 = 5,270 words). Descriptive data are located in Table 2.

Table 2.

Database Descriptive Statistics for United States Supreme Court Justices.

Justice	Speeches	Total words	Word count		Imperative speech		Self-references
Justice	Speeches	Total words	M	SD	M(%)	SD(%)	M(%)	SD(%)
Thomas	103	352,200	3,419.42	2,891.43	0.29	0.24	0.88	0.66
Breyer	95	401,655	4,227.95	4,047.44	0.31	0.18	0.80	0.44
Alito	76	305,501	4,019.75	3,173.77	0.37	0.29	0.72	0.68
Scalia	71	239,734	3,376.54	2,866.30	0.31	0.17	0.71	0.69
Stevens	63	233,183	3,701.32	3,598.68	0.29	0.18	0.64	0.32
Ginsburg	62	221,758	3,576.74	3,255.08	0.27	0.20	0.66	0.42
Sotomayor	56	280,721	5,012.88	3,578.27	0.28	0.15	0.59	0.43
Roberts	43	171,350	3,984.88	2,028.44	0.31	0.17	0.44	0.19
Kennedy	23	90,289	3,925.61	2,575.03	0.37	0.19	0.50	0.64
Souter	22	76,720	3,487.27	2,977.85	0.26	0.17	0.88	0.82
Kagan	17	96,356	5,668.00	2,743.17	0.27	0.11	0.52	0.22
Gorsuch	13	44,076	3,390.46	1,716.70	0.39	0.19	0.37	0.17

Note. Data are sorted in descending order of speech frequency (total number of dissenting opinions per justice). Imperative speech was calculated using the nine-word imperative dictionary.

Consistent with Study 2, we evaluated the connection between scores on our imperative dictionary (M = 0.31%, SD = 0.20%; Q1 = 0.19%, Mdn = 0.28%, Q3 = 0.39%) and now, the rate of “I” pronouns (M = 0.70%, SD = 0.55%; Q1 = 0.40%, Mdn = 0.56%, Q3 = 0.78%). Note, the base rates for imperatives and self-references are low because of the relatively high word counts in dissenting opinions.

Results

We used linear mixed models to evaluate the prior relationship with several controls as random effects. First, we controlled for dissenting opinion author since there are multiple nonindependent observations from 12 Supreme Court justices. Second, we controlled for case, as multiple authors can write dissenting opinions about the same topic. Finally, we controlled for year as a fixed effect to account for conventions that may change in the court over time. This variable was entered as a fixed effect to aid in model convergence. Independent and dependent variables were natural log-transformed to reduce data skewness according to the Study 2a procedure and then standardized.

Imperatives and “I” pronouns were negatively associated in our sample of Supreme Court dissents, β = −0.09, SE = 0.04, t = −2.24, p = .026, R²c = 0.14. These data support our second hypothesis, which suggests that responding to an imperative is associated with a reduction in self-focus. Visual inspection of the correlation coefficients for each justice, however, revealed that Justices Kennedy, Souter, Ginsburg, and Scalia demonstrate a clear negative association between imperatives and “I” pronouns (right panel of Figure 1). Several other justices showed this pattern in the predicted direction but failed to reach statistical significance, likely due to the relatively small number of texts for each individual author.

Exploratory Analyses

Exploratory data suggest that imperatives were positively associated with “we” pronouns, β = 0.08, SE = 0.04, t = 2.20, p = .028, R²c = 0.30, and impersonal pronouns, β = 0.18, SE = 0.04, t = 4.59, p < .001, R²adj = 0.05, but negatively associated with “you” pronouns, β = −0.10, SE = 0.04, t = −2.69, p = .007, R²c = 0.26.

Discussion

Despite some mixed findings for the exploratory results between Study 3 and Study 4, the evidence in the present study generally suggests that responding to imperatives associates with a reduced self-focus and a reduced “you”-focus. Consistent with Hypothesis 2, these pronoun patterns indicate a reduction in “individuated identity” for the imperative decoders compared with the nonimperative decoders (Simmons et al., 2005, p. 933). Therefore, after people internalize that the world must be a particular way, they often express fewer self-references and other indicators of individuation. While “she/he” pronouns were unassociated with imperative speech in Study 4, the relationship formed a trend in the direction consistent with Study 3.

General Discussion

The data from four studies suggest that communicating (encoding) and responding to (decoding) imperatives are associated with different psychological dynamics. Our a priori and exploratory language analyses provide evidence that communicating an imperative requires the inclusion of collective references (“we” pronouns) to plausibly amplify speaker credibility and “they” pronouns to establish degrees of difference between in-groups and out-groups (Bäck et al., 2018; Gustafsson Sendén et al., 2014; Perdue et al., 1990). On the other hand, people who respond to an imperative face psychological costs with a reduced singular and self-focus relative to those who do not respond to an imperative. The analyses used large-scale experimental and observational studies to provide a holistic evaluation of how imperatives affect and relate to verbal patterns by communicators and respondents.

The prior evidence marks a shift in how people orchestrate and make sense of imperatives as communicators or respondents. Consistent with the idea that imperatives require social and psychological support to be communicated, but might undermine or cognitively constrain the autonomy of speakers, these language effects are some of the first to document the bidirectional impact of imperative speech.

It is important to highlight the subtleties of our experimental manipulations that led to differences in how people wrote, thought, and felt during the imperative process. For example, in Study 1, telling people to defend “why America must protect itself” rather than describe “why America benefits from protecting itself” is a small verbal difference with downstream consequences for the psychology of communicators. Our manipulations are similar to those in mindset research (Dweck, 2019), which suggests that how people approach a problem and their prior beliefs can affect their appraisal of an issue. For example, people who believe that self-control is a nonlimited resource do not show willpower deficiencies compared with people who believe that self-control is a limited resource (Job et al., 2010). Furthermore, appraising a temptation (a marshmallow) as a chewy, sweet treat versus a round, white moon leads to changes in how well children can resist temptation (Mischel & Baker, 1975). Our evidence suggests that how people think about an issue—either through an imperative frame or nonimperative frame—affects their writing style and their psychology.

The manipulations in Study 3 asked people to describe why they agreed or disagreed with a university policy that students “must have exceptional scores” for admission or “are encouraged to have exceptional scores” for admission. We underscore the simplicity of this manipulation as well because it changed how writers perceived their ability to place the self in an opinion. This evidence is consistent with Bénabou et al. (2018) who suggest that imperatives can act as cognitive straightjackets and constrain how people think and feel about a particular issue. Our experimental and field data suggest that imperatives can limit a respondent’s willingness to put themselves in a communication act. Since most societies value democratic ideals such as free speech and freedom of expression, it is important to recognize potential blind-spots and when these ideals can become intentionally or unintentionally undermined with words.

Empirical Contributions

We believe that there are five key contributions of this article. The first contribution is methodological, as we developed a dictionary of imperative speech based on modal auxiliary verbs, which can be used and expanded on by other researchers. Second, our immediacy and psychological distancing effects are consistent inside and outside the laboratory, and across countries (Study 2b). Therefore, we suggest that the relationship between imperatives and psychological distancing represents a rich psychological phenomenon, instead of an artifact of region or time period. Third, classic psycholinguistics and speech act theory research has evaluated characteristics of imperatives and their functional value (Austin, 1975; Clark, 1996), but often ignored their psychological correlates as indicated by styles words. Our studies describe the psychological conditions that result from high and low rates of imperatives, namely when a leader seeks to sanction or request an action, amplify their credibility, or when people respond to an imperative. Therefore, our findings extend speech act theory by providing evidence that imperatives are supported and revealed through small style words. Content matters for imperative speech, but style can inform psychological experiences when communicating and responding to imperatives as well.

Relatedly, our fourth contribution considers how imperatives affect communicators differently than respondents. We call this the interpersonal imperatives asymmetry, a linguistic phenomenon where the psychological requirements and consequences associated with imperatives are different for encoders than decoders. Imperatives are purposeful and the communicator likely wants to control the reaction of the respondent. Therefore, it may be important to educate listeners in cognitive mechanisms to protect against the adverse effects of imperatives. Procedures can be instituted that may structurally offset the negative impact of imperatives, perhaps by formally requiring active, analytic thinking.

Finally, our exploratory analyses revealed new and unexpected relationships between imperatives, encoding and decoding processes, and pronouns. Chief among them is the rate of “they” pronouns in the encoding process, which were positively associated with imperatives. This evidence is consistent with prior work that suggests people of high-social rank tend to use collective pronouns at higher rates relative to people of low-social rank (Kacewicz et al., 2014), and high rates of “they” pronouns might indicate increased intergroup bias to separate one collective from another (Gustafsson Sendén et al., 2014; Perdue et al., 1990). Despite these data, it is unclear if imperative speech facilitates feelings of high status and collective orientation, or if status affects how people communicate imperatives and pronouns. Future work should use experiments to clarify the relationship between status, imperative speech, and multiple pronoun types. Such exploratory findings highlight the importance of pairing a priori, theory-driven predictions with exploratory analyses to discover novel and unanticipated relationships between language variables (e.g., Markowitz, 2018, 2019; Shah et al., 2015; van Atteveldt & Peng, 2018).

Limitations and Future Directions

As noted, our imperative dictionary contained nine function words that have imperative characteristics, though other words could be considered in future versions of the dictionary (e.g., can, do) and imperatives exist in other forms through content words. We drew on the established LIWC2015 discrepancy category for our imperative dictionary, though future research should use other approaches to capture imperative statements beyond function words, and expand the feature set to include other dimensions that may relate to communicating or responding to imperatives. A second limitation suggests that our effect sizes were relatively small and required large samples, especially in the field studies. Effect sizes for language patterns are often small in experiments (Kramer et al., 2014) and field studies (Markowitz, 2019; Pennebaker et al., 2014), and detectable patterns likely benefited from scale. Therefore, we encourage future research to continue with large scale, a priori powered studies to make grounded predictions about how language reflects social and psychological processes.

Third, our field studies evaluated speakers with high rank (e.g., politicians, presidents, Supreme Court justices) and we observed that their communication patterns reflect the interpersonal imperatives asymmetry. Future research should expand the populations of interest to investigate if these language patterns hold across different types of speakers and on different communication platforms (e.g., social networking sites).

Fourth, our study using U.S. Supreme Court statements (Study 4) suggests that dissenting opinions with high rates of imperatives also contain low rates of self-references. We acknowledge that a stricter test of this idea might compare the writing style of court majority opinions (how the court decided) versus dissents to evaluate “I” pronoun and imperative connections. We decided against this approach since the results would likely capture speaking style effects of the justices, instead of speaking style effects related to imperative speech. Since opinions and dissents are not written by the same person for the same case, this was a nontrivial possibility. After qualitatively reviewing a selection of dissents, however, we observed that imperatives often accompany and directly refer to court opinion matters, such as the example from Justice Thomas’s dissent in Presley v. Georgia (2010):

The statement that a “trial court must consider reasonable alternatives to closing the proceeding” ibid. (quoting Waller, supra, at 48), does not definitively establish who must suggest alternatives to closure that the trial court must then consider, nor does it expressly address whether the trial court must suggest such alternatives in the absence of a proffer. I concede that the language can easily be read to imply the latter, and the Court may well be right that a trial court violates the Sixth Amendment if it closes the courtroom without sua sponte considering reasonable alternatives to closure.

Based on our empirical concerns about measuring speaker effects and our qualitative review that suggests imperatives in dissents often refer to court material, we felt that it was appropriate to use dissenting opinions as the text of interest in Study 4.

We encourage future investigations to explore how different manipulations in our experiments, particularly related to social identity and in-group processes, might affect the results. For example, it might be important to examine how American participants would give a speech if they were asked to speak in another country about why a particular social policy must change. Our studies in this article examined how speakers from one country (e.g., the United States, the Soviet Union) were communicating to an audience in largely the same country. We believe that investigating the overlap between imperatives, social identity, in-group and out-group dynamics, and language patterns is an important line of research.

We also believe that future research should also investigate the decoding process with a mixed methods approach. Our studies observed that responding to an imperative created psychological distancing through “I” pronouns, though it is also possible that our manipulation in Study 3 might have created a perceived difference in formality (e.g., participants in the imperative condition may have perceived the process as more formal than participants in the nonimperative condition) or threat (e.g., suggesting that someone must have exceptional grades could be perceived as threatening compared with being encouraged to have exceptional grades). Assessing this possibility directly with language is a challenge and therefore, future work would benefit from incorporating qualitative techniques to evaluate how people perceive others or institutions that communicate in an imperative or nonimperative manner.

Finally, a common perspective on imperatives is that they sanction, request, or encourage an action of some kind. The aim of this article was to measure the psychological correlates of imperative speech through function words and our work suggests that responding to an imperative inhibits self-focus. It is unclear if this phenomenon also extends to inhibiting action, such as voting or donating resources. The International Federation of Red Cross and Red Crescent Societies (2019) assists victims of disasters and often advertises that “We need your help.” Are people less likely to donate to the Red Cross after an imperative statement relative to an alternative (e.g., “Please help us”)? Such questions can be addressed with additional experimental data and are crucial to investigate how imperatives might be associated with a reduction in behavioral intention or action.

Conclusion

This article provides evidence of an interpersonal imperatives asymmetry, a phenomenon that suggests communicating an imperative requires psychological support through high rates of collective pronouns. However, imperatives can undermine self-focus after an imperative prompt compared with a nonimperative prompt for respondents. These patterns were observed in experimental and field studies, in conjunction with exploratory analyses, to suggest that word patterns associate with social and psychological experiences of communicators. The effects extend speech act theory and other models that assume imperative speech can be characterized by communication content. Function words, particularly pronouns, hold rich psychological value and indicate imperative dynamics for communicators and respondents.

We believe these results have important interpersonal implications in many settings, such as politics, organizations, and interactions where potential power differences are apparent. For example, these data can inform how politicians need to communicate directives to the public and the potential impact on those who hear orders from their representatives; bosses who direct employees to perform an action and the potential lack of agency employees might feel as a result of the directive; parents who tell children to perform an action using imperatives and how children might feel after hearing an imperative. These examples demonstrate that imperatives are common, nontrivial, and functional, while the current article reveals that they are also psychologically rich in the encoding and decoding process.

Footnotes

Acknowledgements

We would like to thank three anonymous reviewers and the editor for their thoughtful improvements related to this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

David M. Markowitz

Notes

Author Biographies

David M. Markowitz is an assistant professor in the School of Journalism and Communication at the University of Oregon. He uses language data from natural repositories to make inferences about what people are thinking, feeling, and experiencing psychologically. His work has appeared in outlets such as the Journal of Communication, Communication Research, and Journal of Language and Social Psychology. He received his PhD from Stanford University and his undergraduate and master’s degrees from Cornell University.

Paul Slovic is a founder and president of Decision Research and professor of psychology at the University of Oregon, who studies human judgment, decision making, and risk analysis. He publishes extensively and serves as a consultant to industry and government. He has received honorary doctorates from the Stockholm School of Economics (1996) and the University of East Anglia (2005). He was elected to the American Academy of Arts and Sciences in 2015 and the National Academy of Sciences in 2016.

References

Austin

J. L.

(1975). How to do things with words. Harvard University Press.

Bäck

E. A.

Bäck

Sendén

M. G.

Sikström

(2018). From I to we: Group formation and linguistic adaption in an online xenophobic forum. Journal of Social and Political Psychology, 6(1), 76-91. https://doi.org/10.5964/jspp.v6i1.741

Bauer-Wolf

(2019). Young people support free college. https://www.insidehighered.com/news/2019/05/03/poll-support-free-college-among-young-people

Bénabou

Falk

Tirole

(2018, July). Narratives, imperatives, and moral reasoning (Working Paper 24798). https://doi.org/10.3386/w24798

Blackburn

K. G.

Yilmaz

Boyd

R. L.

(2018). Food for thought: Exploring how people think and talk about food online. Appetite, 123, 390-401. https://doi.org/https://doi.org/10.1016/j.appet.2018.01.022

Boothby

E. J.

Clark

M. S.

Bargh

J. A.

(2014). Shared experiences are amplified. Psychological Science, 25(12), 2209-2216. https://doi.org/10.1177/0956797614551162

Boyd

R. L.

(2018). Mental profile mapping: A psychological single-candidate authorship attribution method. PLOS ONE, 13(7), Article e0200588. https://doi.org/10.1371/journal.pone.0200588

Boyd

R. L.

Pennebaker

J. W.

(2015). A way with words: Using language for psychological science in the modern era. In Dimofte

Haugtvedt

Yalch

(Eds.), Consumer psychology in a social media world (pp. 222-236). Routledge.

Bull

Fetzer

(2006). Who are we and who are you? The strategic use of forms of address in political interviews. Text & Talk: An Interdisciplinary Journal of Language, Discourse Communication Studies, 26(1), 3-37. https://doi.org/10.1515/TEXT.2006.002

10.

Carbone

(1975). Stylistic variables as related to source credibility: A content analysis approach. Speech Monographs, 42(2), 99-106. https://doi.org/10.1080/03637757509375884

11.

Carter

(2017, September 12). Voters need a digital defense. The New York Times. https://www.nytimes.com/2017/09/12/opinion/jimmy-carter-electronic-voting.html

12.

Chung

Pennebaker

J. W.

(2007). The psychological functions of function words. In Fiedler

(Ed.), Social communication (pp. 343-359). Psychology Press.

13.

Clark

H. H.

(1996). Using language. Cambridge University Press.

14.

Cohn

M. A.

Mehl

M. R.

Pennebaker

J. W.

(2004). Linguistic markers of psychological change surrounding September 11, 2001. Psychological Science, 15(10), 687-693. https://doi.org/10.2307/40064028

15.

D’Andrade

R. G.

Wish

(1985). Speech act theory in quantitative research on interpersonal behavior. Discourse Processes, 8(2), 229-259. https://doi.org/10.1080/01638538509544615

16.

De Fina

. (1995). Pronominal choice, identity, and solidarity in political discourse. Text: Interdisciplinary Journal for the Study of Discourse, 15(3), 379-410. https://doi.org/10.1515/text.1.1995.15.3.379

17.

Dweck

(2019). Mindset: The new psychology of success. Ballantine Books.

18.

Gardner

W. L.

Gabriel

Lee

A. Y.

(1999). “I” value freedom, but “we” value relationships: Self-construal priming mirrors cultural differences in judgment. Psychological Science, 10(4), 321-326. https://doi.org/10.1111/1467-9280.00162

19.

Golder

S. A.

Macy

M. W.

(2011). Diurnal and seasonal mood vary with work, sleep, and daylength across diverse cultures. Science, 333(6051), 1878-1881. https://doi.org/10.1126/science.1202775

20.

Graham

Haidt

Nosek

B. A.

(2009). Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology, 96(5), 1029-1046. https://doi.org/10.1037/a0015141

21.

Gustafsson Sendén

Lindholm

Sikström

. (2014). Selection bias in choice of words: Evaluations of “I” and “we” differ between contexts, but “they” are always worse. Journal of Language and Social Psychology, 33(1), 49-67. https://doi.org/10.1177/0261927X13495856

22.

Hart

R. P.

Childers

J. P.

(2004). Verbal certainty in American politics: An overview and extension. Presidential Studies Quarterly, 34(3), 516-535. https://doi.org/10.1111/j.1741-5705.2004.00210.x

23.

Hauch

Blandón-Gitlin

Masip

Sporer

S. L.

(2015). Are computers effective lie detectors? A meta-analysis of linguistic cues to deception. Personality and Social Psychology Review, 19(4), 307-342. https://doi.org/10.1177/1088868314556539

24.

Holtzman

N. S.

Tackman

A. M.

Carey

A. L.

Brucks

M. S.

Küfner

A. C. P.

Deters

F. G.

Back

M. D.

Donnellan

M. B.

Pennebaker

J. W.

Sherman

R. A.

Mehl

M. R.

(2019). Linguistic markers of grandiose narcissism: A LIWC analysis of 15 samples. Journal of Language and Social Psychology, 38(5-6), 773-786. https://doi.org/10.1177/0261927X19871084

25.

Hyland

(2003). Self-citation and self-reference: Credibility and promotion in academic publication. Journal of the American Society for Information Science and Technology, 54(3), 251-259. https://doi.org/10.1002/asi.10204

26.

International Federation of Red Cross and Red Crescent Societies. (2019). We need your help. https://media.ifrc.org/ifrc/donate/

27.

Ireland

M. E.

Mehl

(2014). Natural language use as a marker of personality. In Holtgraves

T. M.

(Ed.), The Oxford handbook of language and social psychology (pp. 201-218). Oxford University Press.

28.

Job

Dweck

C. S.

Walton

G. M.

(2010). Ego depletion—is it all in your head? Implicit theories about willpower affect self-regulation. Psychological Science, 21(11), 1686-1693. https://doi.org/10.1177/0956797610384745

29.

Johnson

(1994). Gender, legitimate authority, and leader-subordinate conversations. American Sociological Review, 59(1), 122-135. https://doi.org/10.2307/2096136

30.

Jordan

K. N.

Sterling

Pennebaker

J. W.

Boyd

R. L.

(2019). Examining long-term trends in politics and culture through language of political leaders and cultural institutions. Proceedings of the National Academy of Sciences of the United States of America, 116(9), 3476-3481. https://doi.org/10.1073/pnas.1811987116

31.

Kacewicz

Pennebaker

J. W.

Davis

Jeon

Graesser

A. C.

(2014). Pronoun use reflects standings in social hierarchies. Journal of Language and Social Psychology, 33(2), 125-143. https://doi.org/10.1177/0261927x13502654

32.

Kramer

A. D. I.

Guillory

J. E.

Hancock

J. T.

(2014). Experimental evidence of massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences, 111(24), 8788-8790. https://doi.org/10.1073/pnas.1320040111

33.

Larrimore

Jiang

Larrimore

Markowitz

Gorski

(2011). Peer to peer lending: The relationship between language features, trustworthiness, and persuasion success. Journal of Applied Communication Research, 39(1), 19-37. https://doi.org/10.1080/00909882.2010.536844

34.

Liehr

Mehl

M. R.

Summers

L. C.

Pennebaker

J. W.

(2004). Connecting with others in the midst of stressful upheaval on September 11, 2001. Applied Nursing Research, 17(1), 2-9. https://doi.org/10.1016/J.APNR.2003.10.010

35.

Mairesse

Walker

M. A.

Mehl

M. R.

Moore

R. K.

(2007). Using linguistic cues for the automatic recognition of personality in conversation and text. Journal of Artificial Intelligence Research, 30(1), 457-500. https://dl.acm.org/citation.cfm?id=1622649

36.

Markowitz

D. M.

(2018). Academy Awards speeches reflect social status, cinematic roles, and winning expectations. Journal of Language and Social Psychology, 37(3), 376-387. https://doi.org/10.1177/0261927x17751012

37.

Markowitz

D. M.

(2019). What words are worth: National Science Foundation grant abstracts indicate award funding. Journal of Language and Social Psychology, 38(3), 264-282. https://doi.org/10.1177/0261927X18824859

38.

Markowitz

D. M.

Griffin

D. J.

(2020). When context matters: How false, truthful, and genre-related communication styles are revealed in language. Psychology, Crime & Law, 26, 287-310. https://doi.org/10.1080/1068316X.2019.1652751

39.

Markowitz

D. M.

Hancock

J. T.

(2014). Linguistic traces of a scientific fraud: The case of Diederik Stapel. PLOS ONE, 9(8), Article e105937. https://doi.org/10.1371/journal.pone.0105937

40.

Markowitz

D. M.

Hancock

J. T.

(2016). Linguistic obfuscation in fraudulent science. Journal of Language and Social Psychology, 35(4), 435-445. https://doi.org/10.1177/0261927X15614605

41.

Markus

H. R.

Kitayama

(1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98(2), 224-253. https://doi.org/10.1037/0033-295X.98.2.224

42.

Mazarr

(2016). Rethinking risk in national security: Lessons of the financial crisis for risk management. Palgrave Macmillan.

43.

Mischel

Baker

(1975). Cognitive appraisals and transformations in delay behavior. Journal of Personality and Social Psychology, 31(2), 254-261. https://doi.org/10.1037/h0076272

44.

Moldovan

Rus

Graesser

A. C.

(2011). Automated speech act classification for online chat. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.231.5461&rep=rep1&type=pdf

45.

Nave

N. N.

Shifman

Tenenboim-Weinblatt

(2018). Talking it personally: Features of successful political posts on Facebook. Social Media + Society, 4(3), Article e8477. https://doi.org/10.1177/2056305118784771

46.

Obama

(2007). Obama calls for elimination of subsidies to student loan providers. https://www.presidency.ucsb.edu/documents/press-release-obama-calls-for-elimination-subsidies-student-loan-providers

47.

Pennebaker

J. W.

(2011). The secret life of pronouns: What our words say about us. Bloomsbury Press.

48.

Pennebaker

J. W.

Booth

R. J.

Boyd

R. L.

Francis

M. E.

(2015). Linguistic Inquiry and Word Count: LIWC2015. Pennebaker Conglomerates.

49.

Pennebaker

J. W.

Chung

C. K.

Frazee

Lavergne

G. M.

Beaver

D. I.

(2014). When small words foretell academic success: The case of college admissions essays. PLOS ONE, 9(12), Article e115844. https://doi.org/10.1371/journal.pone.0115844

50.

Pennebaker

J. W.

Stone

L. D.

(2003). Words of wisdom: Language use over the life span. Journal of Personality and Social Psychology, 85(2), 291-301. https://doi.org/10.1037/0022-3514.85.2.291

51.

Perdue

C. W.

Dovidio

J. F.

Gurtman

M. B.

Tyler

R. B.

(1990). Us and them: Social categorization and the process of intergroup bias. Journal of Personality and Social Psychology, 59(3), 475-486. https://doi.org/10.1037/0022-3514.59.3.475

52.

Petriglieri

J. L.

(2011). Under threat: Responses to and the consequences of threats to individuals’ identities. Academy of Management Review, 36(4), 641-662. https://doi.org/10.5465/amr.2009.0087

53.

Presley v. Georgia, No. 09-5270 (2010). https://caselaw.findlaw.com/us-supreme-court/09-5270.html

54.

Proctor

L. I.-W.

(2011). The 1st person plural in political discourse—American politicians in interviews and in a debate. Journal of Pragmatics, 43(13), 3251-3266. https://doi.org/10.1016/J.PRAGMA.2011.06.010

55.

Rude

Gortner

E.-M.

Pennebaker

(2004). Language use of depressed and depression-vulnerable college students. Cognition and Emotion, 18(8), 1121-1133. https://doi.org/10.1080/02699930441000030

56.

Salaita

(2005). Ethnic identity and imperative patriotism: Arab Americans before and after 9/11. College Literature, 32(2), 146-168. https://www.jstor.org/stable/25115271?seq=1#metadata_info_tab_contents

57.

Salganik

(2017). Bit by bit: Social research in the digital age. Princeton University Press.

58.

Shah

D. V

Cappella

J. N.

Neuman

W. R.

(2015). Big data, digital media, and computational social science: Possibilities and perils. ANNALS of the American Academy of Political and Social Science, 659(1), 6-13. https://doi.org/10.1177/0002716215572084

59.

Simmons

R. A.

Gordon

P. C.

Chambless

D. L.

(2005). Pronouns in marital interaction. Psychological Science, 16(12), 932-936. https://doi.org/10.1111/j.1467-9280.2005.01639.x

60.

Tackman

A. M.

Sbarra

D. A.

Carey

A. L.

Donnellan

M. B.

Horn

A. B.

Holtzman

N. S.

Edwards

T. S.

Pennebaker

J. W.

Mehl

M. R.

(2019). Depression, negative emotionality, and self-referential language: A multi-lab, multi-measure, and multi-language-task research synthesis. Journal of Personality and Social Psychology, 116(5), 817-834. https://doi.org/10.1037/pspp0000187

61.

Tausczik

Y. R.

Pennebaker

J. W.

(2010). The psychological meaning of words: LIWC and computerized text analysis methods. Journal of Language and Social Psychology, 29(1), 24-54. https://doi.org/10.1177/0261927X09351676

62.

Trafimow

Triandis

H. C.

Goto

S. G.

(1991). Some tests of the distinction between the private self and the collective self. Journal of Personality and Social Psychology, 60(5), 649-655. https://doi.org/10.1037/0022-3514.60.5.649

63.

van Atteveldt

Peng

T.-Q

. (2018). When communication meets computation: Opportunities, challenges, and pitfalls in computational communication science. Commu-nication Methods and Measures, 12(2-3), 81-92. https://doi.org/10.1080/19312458.2018.1458084

64.

Van Swol

L. M.

Carlson

C. L

. (2015). Language use and influence among minority, majority, and homogeneous group members. Communication Research, 44(4), 512-529. https://doi.org/10.1177/0093650215570658

65.

Weintraub

(1981). Verbal behavior: Adaptation and psychopathology. Springer.

66.

Wiener

Mehrabian

(1968). Language within language: Immediacy, a channel in verbal communication. Appleton-Centry-Crofts.