The effects of modal value and imperative mood on self-predicted compliance to health guidance: the case of COVID-19

3.1.1 Closed-questions

The participants were presented with either “Group A” or “Group B” images (Table 2), and were asked “How likely or unlikely would you be to follow the guidance in this public health message if such measures were re-introduced as a result of a new COVID-19 variant?”. They responded through a seven-point Likert scale: “extremely likely”, “very likely”, “fairly likely”, “neither likely nor unlikely”, “fairly unlikely”, “very unlikely”, “extremely unlikely” (presented at random in either forward or reverse order); a further “don’t know” response was anchored at the end. The images were selected at random per sample group and were identical except for the changes in modality or mood.

Modality was expressed through modal verbs and varied in value only (see analytical framework, Section 3.2.1): for each pair of images, one message would show a medium modality value of obligation or probability (“should”, “could”), and the other a high value, connoting strong imposition or certainty (“must”, “will”) (posters 1–3, Table 2, corresponding to fear appeals, moralising and self-efficacy messages, three message types employed in UK official campaigns). The effects of mood were examined by displaying the same directive in indicative and imperative mood (poster 4, Table 2, corresponding to a self-efficacy message).

3.1.2 Open-questions

The participants were presented with Figures 1 and 2 and asked “Looking at the public health communication below, what do you think it is trying to say or get across? And what is your reaction to this? Please write in below.” Both posters formulate the command in imperative mood, but differ in message type: Figure 1 is a fear appeal (the course of action is justified by evoking fear, “we’re all at risk”); Figure 2 is moralising (the course of action is evoked by appealing to moral values) – two message types with mixed-assessments in the literature (Berry 2006; Täuber 2018).

The interaction between message type and formulation of directive (imperative) was examined considering the participants’ engagement with the directive, exploring the inclusion of the course of action advised in the participants’ responses, and comparing the wording used by the participants in reporting the health guidance with the original formulation. The analysis assumed that, the more references the participants make to the guidance, the easier it is for them to focus on the behaviour promoted, thus increasing the likelihood of compliance.

3.2.1 Modality

Modality encodes speakers’ judgements or requests. SFL distinguishes “Modalisation” and “Modulation”, depending on whether what is expressed constitutes a proposition (a description, Modalisation), or a proposal (Modulation). Modalisation involves expressions of Probability (“You can get ill”) and Usuality (“He usually gets travel-sick”) (Halliday and Matthiessen 2004: 147). Modulation involves Obligation (“You should wash your hands”), in command speech function, and Inclination (“I will wash my hands”), in offer speech function (Halliday and Matthiessen 2004: 147). Modality evaluations have polarity (positive or negative) and may show high, medium or low values (e.g., must, should, can) (Halliday and Matthiessen 2004: 143–149). Depending on the speaker’s viewpoint, modality has a subjective orientation, if the speaker stands as the source of the evaluation (“I oblige you to wash your hands”), or an objective orientation (“It is compulsory …”), if the evaluation source is presented as objective (Halliday and Matthiessen 2004: 149–150). The source of the evaluation can be explicit (i.e., “I oblige …” [subjective:explicit], “It is certain …” [objective:explicit]), or implicit (“You must wash …”, [subjective:implicit], “Certainly …”, [objective:implicit]) (2004: 149–150). See Figure 3 for a summary.

Figure 3:

Modality as understood in SFL (adapted from Halliday and Matthiessen 2004: 150, Figure 4.25).

3.2.2 Mood

The mood system is the most basic grammatical encoding of the interpersonal function and defines the grammaticalization of the primary speech functions (statements, questions, offers and commands) (Halliday and Matthiessen 2004: 107–108). As a clause element, the Mood is composed of the Finite and the Subject elements. The Finite is part of a verbal group and it expresses tense (e.g., do, has) or modality (e.g., can, might); the Subject is a nominal group or personal pronoun and corresponds to the grammatical subject (Halliday and Matthiessen 2004: 111–112). Realisation variations in the inclusion and ordering of the Finite and Subject (Figure 4, represented by the arrows) define two main mood types: the imperative and the indicative, the latter further subdivided into declarative and interrogative (Halliday and Matthiessen 2004: 114).

Figure 4:

Overview of the mood system as understood in SFG (adapted from Halliday and Matthiessen 2004: 23, Figure 1.9).

The indicative mood is typically used in information exchanges and presents both elements: SubjectˆFinite for declaratives (“He [S] must [F] be ill”), the unmarked^[1] structure of statements; FiniteˆSubject for yes/no interrogatives (“Is [F] he [S] ill?”), and Wh-ˆFinite for Wh- interrogatives (Who is [F] ill?), the unmarked structures of questions.

The imperative mood is typically used for commands; its unmarked positive realisation does not present any mood element (“Wash your hands”), but we can find the Finite in negative imperatives (“Don’t …!) or to add emphasis (“Do wash …”), and sometimes also the Subject (“You wash …!”) (Halliday and Matthiessen 2004: 138). Besides these congruent realisations of speech functions, these can also be realised noncongruently (interpersonal metaphors) – e.g., commands realised as yes/no interrogatives (“Can you wash …?”) or declaratives (“you should wash …”), which offer more potential for negotiation (2004: 628, 632).

4.1.1 Direct commands with modal expressions of obligation

Messages 1 and 2 (posters 1 and 2, Table 2) constitute self-efficacy and moralising messages respectively:

The messages indicate the guidance with direct commands and where it is to be followed (hospitals, public transport). Message 2 includes the rationale for following the guidance (protection). The modal auxiliaries “should” and “must” involve an implicit subjective orientation, and only differ in value (medium, high).

Figures 5 and 6^[2] show that self-predicted compliance is high for both modal values in the two message types, thus challenging the expectation of obligation strength (modal value) functioning as a determining factor of public adherence to guidance.

Figure 5:

Overall self-predicted compliance for high and medium values of Modulation:Obligation in message 1 (self-efficacy).

Figure 6:

Overall self-predicted compliance for high and medium values of Modulation:Obligation in message 2 (moralising).

For message 1 (self-efficacy), there is a slightly significant difference in self-predicted compliance per Group A (“should”) (M = 5.40, SD = 1.712)^[3] and Group B (“must”) (M = 5.13, SD = 1.860); t (1,061) = 2.201, p = 0.028. Overall compliance for the self-efficacy message (Figure 5) is somewhat higher for directives with medium value modality (“should”). While these results would indicate a disinclination for high values of obligation, suggesting a general reluctance towards directives that may appear more face-threatening, for message 2 (moralising) the difference in self-predicted compliance per Group A (“should”) (M = 5.27, SD = 1.761) and Group B (“must”) (M = 5.22, SD = 1.808) is not significant; t (1,063) = 0.511, p = 0.610 (see also Figure 6).

These results could be partly affected by the composition of message 2 poster: the modalised command is presented in smaller print, allowing the participants to read it as an elaboration of the visually salient guidance in imperative, displayed in larger yellow fonts (Kress and van Leeuwen 2021: 210–211).

Tables 3 and 4 below report the overall self-predicted compliance per age group for the two message types, and identify the older groups (55–75 and 45–54 years) as those who show the highest compliance for both messages in the four variations. However, preference for higher or medium modality values varies by age group and message type. For the self-efficacy message (Table 3), the participants favour medium modality values (“should”) regardless of age. While differences in modality preferences are minimal for the older groups, the younger participants, particularly the 25–34 age group, report a strong disinclination for high values of obligation (“must”), alongside the lowest levels of self-predicted compliance (Table 3).

Instead, for message 2 (moralising), the older populations (age groups 45–54 and 55–75) and the 25–34 age group favour the high imposition of “must” (Table 4), while the youngest age group (16–24) shows the highest rejection for strong obligation (see Appendix, Figure 9, for diagram with significant mean differences of self-predicted compliance per group and framing variation).

The self-efficacy message triggered high levels of self-predicted compliance across all age groups except the 25–34 group; instead, the moralising message worked better for the older audiences, with the younger populations showing medium degrees of compliance (Table 4). The tenor clash observed in message 2b, with the directive in imperative formulated as a request (“please wear …”) and the command connoting strong obligation (“You must …”), could contribute to the resistance of the young respondents.

4.1.2 Indirect commands: compliance conveyed from the consequences of the health risk

Fear appeals, like message 3, are common health messages (Berry 2006: 109). In message 3, the need to follow the guidance is inferred from the negative consequences of non-compliance, and the likelihood of risk is modalised. The visuals accompanying the text further emphasise the severity of the virus (poster 3, Table 2): a long shot of an elderly man in bed receiving assistance acquires visual salience through colours (bright orange and yellow), the absence of gaze contact with the viewer contributes to offer the represented scene, in impersonal distance, as factual information (Kress and van Leeuwen 2021: 118).

Although compliance values are high for both variations of the message (Figure 7), there is a significant difference in self-predicted compliance per Group A (“will”) (M = 5.40, SD = 1.684) and Group B (“could”) (M = 5.13, SD = 1.810); t (1,058) = 2.532, p = 0.011.

Figure 7:

Overall self-reported compliance for high and medium values of Modalisation:Probability in fear appeals (message 3).

The association of stronger modality (“will”) with greater compliance suggests that conveying certainty when reporting the outcomes of a health risk promotes the audience’s inclination to engage in preventive actions, and reflect previous studies, which stress the importance of informing the audience of the magnitude of the health threat in promoting guidance (e.g., Berry 2006: 109; Hale and Dillard 1995).

Contrary to the modulation of obligation in self-efficacy or moralising messages, Table 5 shows that all age groups reported a preference for high modality values for the fear appeal.

The 25–34 age group showed the lowest compliance for medium probability (“could”), and the youngest population (16–24) showed the greatest refusal of high probability values (“will”) (see Appendix, Figure 9). As for the other message types, the older audiences reported higher compliance, regardless of the wording. Previous studies have also identified age as a determining factor for fear appeals’ success (Berry 2006: 109; Hale and Dillard 1995; Institute of Medicine 2015).

4.1.3 Health guidance as imperative or declarative

The impact of directness in commands was examined with message 4 (poster 4, Table 2), considering differences in mood. Like message 1, message 4 is a self-efficacy message, but instead of employing modulation of obligation, the directive is either inferred from an information-giving declarative (indirect), or formulated in imperative (direct).

The responses show a significant difference in self-predicted compliance per Group A (declarative) (M = 5.23, SD = 1.819) and Group B (imperative) (M = 5.21, SD = 1.857); t (1,061) = 0.160, p = 0.018 (see Figure 8 for percentages), with higher self-predicted compliance reported for the declarative.

Figure 8:

Overall self-predicted compliance for directives in declarative and imperative mood.

The disinclination for directness is consistent with the preference for medium values of Modulation:Obligation reported for message 1.

Self-predicted compliance for mood variation is conditioned, to a point, by age (Table 6, see also Appendix Figure 9).

While the two oldest age groups reported slightly higher values of compliance for the imperative, the younger audiences, especially the 25–34 age group, reported lower levels for the imperative, supporting the results obtained for Modulation: Obligation in message 1 (Table 3).

4.2.1 Engaging with directives

The effect of direct commands in message reception was examined considering the open question responses for the posters displayed in Figures 1 (fear appeal) and 2 (moralising). Figure 1 precedes the directive (“Follow the rules …”) with the rationale to comply (“We’re all at risk”), attributes salience in the poster (large fonts in yellow) and promotes adherence by emphasising the audience’s lack of immunity. Figure 2 only includes the directive in imperative (“Look her in the eyes and tell …”), and the need to comply with the guidance is inferred from the appeal to the audience’s moral values, triggered by the combination of the textual and the visual messages: an extremely close-up shot of a woman severely ill, explicitly described as “Covid-19 patient” and individuated (“Lorna”). The poster invites engagement and attributes visual salience to the ventilator and the indirect gaze of the patient; the absence of direct eye contact contributes to promote empathy, and the high angle connotes powerlessness (Kress and van Leeuwen 2021: 62, 143).

The responses were coded following the categories noted below. Broadly, we distinguished between those responses that were merely descriptive, of the poster or the respondents’ evaluations, and those responses that included references to the directive.

1. Descriptive

   1. Description: personal evaluation: the participants offer evaluations about the message (“I feel very positive to see this ad.”, “Getting bored with it”). Message fatigue was identified for both posters, especially for Figure 1. The references to the political scandal Partygate^[4] were also coded as evaluations (“This makes me think that there is one rule for the population and one for the prime minister and the MPs […]”).

   2. Description: poster: the participants describe the poster, either commenting on how it looks or the intentions attributed to the messaging (“this poster gets to the point clearly”, “Trying to scare people”).

   3. Description: poster & personal evaluation: the participants provide descriptions of the poster including evaluations (“Pretty clear. Just a great pity that the ones that made the rules didn’t apply them to the,selves (sic) […]”, “Nobody is safe but I don’t think it is necessarily correct”, “Trying to spread alarm. I scoff at such fear porn”). These responses frequently express disagreement, either with the poster or with the government in power at the time.

2. Engagement with the directive

   1. Description: poster & explicit directive: the participants describe the poster and include a directive restating or paraphrasing the guidance (underlined), (“Good visual representation of 2 m social distancing. Tells people that no one is immune from catching covid. Everyone needs to help prevent its spread […]”).

   2. Directive: information: the participants unpack the meaning of the textual or visual information of the poster, but do not explicitly report the directive (“Social distancing helps”, “That ‘bending’ the rules can have serious effects on people’s lives”).

   3. Directive: information & personal evaluation: the participants convey the information provided in the message and expand on it with evaluations, without engaging with the directive (“Telling everyone we (sic) at risk. I don’t care”).

   4. Directive: information & explicit directive: the participants paraphrase the information provided and formulate the directive, either by employing the imperative mood or Modulation:Obligation; “Covid does not discriminate so people need to follow the rules to help keep everyone safe” [Modulation:Obligation:Objective:Implicit],^[5] “everyone is at risk so we must protect each other” [Modulation:Obligation:Subjective:Implicit], “Anyone can catch and spread the illness. Take necessary precautions” [imperative].

   5. Directive: explicit directive: the participants formulate the command, either using the imperative mood or Modulation:Obligation, without reporting information backing the guidance (“Keep your distance […]” [imperative], “Everyone needs to follow the rules […]” [Modulation: Obligation:Objective:Implicit]). The directives formulated as non-finite clauses, with the finite elided, were also coded as “explicit” (“To keep our distance to keep safe”, which stands for “[The health communication is trying to say that we have/need ] to keep our distance to keep safe” [Modulation:Obligation:Objective:Implicit]).

3. Absence of engagement with the health message or the poster

   1. “Don’t know”

   2. No answer

Table 7 provides a summary.

The participants showed reasonable engagement with the directive of the posters. A total of 1,044 responses included some engagement, in contrast to 860 purely descriptive responses, which presented more references to the poster and intentions attributed to message writers than evaluations (Table 7). However, the level of engagement with the directives differed across the messages. Figure 1 provoked a higher engagement, with 635 responses referring to the directive and 313 descriptive responses (Table 7). Instead, Figure 2 is associated with lower engagement, with 409 responses referring to the directive and 547 descriptive responses. Of those participants who engaged with the directive of Figure 2, 110 paraphrased the information provided in the visual format (e.g., “Breaking the rules has serious medical effects”), without including further reference to the guidance. The different engagement with the directives suggests that the participants paid more attention to the guidance promoted in Figure 1 than in Figure 2, which would make Figure 1 a more effective health message. The higher number of descriptive responses triggered by Figure 2 suggests that the visuals and message type (moralising) divert the audience’s attention from the actual guidance; instead, Figure 1 presents a long shot offering the action to be adopted (though the distance is not specified, and the picture is inconsistent with the slogan, “stay at home”).

4.2.2 Reporting the health guidance

The qualitative analysis of the open-questions has identified the imperative mood as the preferred form to report a directive, all directed to a second person addressee which does not include the speaker (Table 7). This preference for the imperative contrasts with the higher self-predicted compliance for the self-efficacy message in declarative mood observed in the quantitative analysis (Figure 8). Such a contrast could be partly influenced by the original adoption of the imperative in the messages considered.

In what follows, we consider the participants’ use of modulation in reporting commands in their answers, observing whether these uses reflect the preferences reported for the closed-questions, and how the participants position themselves towards the guidance. The responses that engage with the directives employing modality are comparatively more prominent for Figure 1 and show a preference for objective implicit forms of modulation, followed by subjective implicit forms (Table 8).

Objective implicit expressions of obligation include “need to”, most commonly (“[…] people need to follow the rules to help keep everyone safe”), “have to” (“Everyone has to follow the rules”), and the relational process “have” with “responsibility” or “duty” as usual attributes (“WE all have responsibility to protect each other”). “Need to” is the expression of obligation most often used to refer to an addressee including the speaker (first person plural) (Table 8). Occasionally, the Subject and the Finite are elided and the participants directly report the hypotactic clause, focusing on the action that needs to be accomplished (“[we have/need] To keep our distance to keep safe”). In these cases, the addressees of the directive have been retrieved from the personal pronouns or possessive determiners included in the response (first person plural determiner “our” in the previous example); when referents were not available, the addressee was noted as “indefinite”. Expressions of obligation with an objective explicit orientation are, in contrast, scarce and only given for Figure 1. The objective explicit orientation construes the obligation as a projection (“It doesn’t matter if you’re relatively young […] therefore it’s important to maintain a safe distance”), and the addressee is often left unspecified, inferred to include both the speaker and third parties.

The most recurrent expression of obligation with a subjective implicit orientation is the medium value modal auxiliary “should”, especially for third person plural addressees (“Everyone is at risk and so everyone should be careful”). References to third person plural addresses may implicitly include the speaker (“everyone” in the example) or not (references to “people”). The high value modal “must” has also been observed, especially when targeting a group of addressees inclusive of the speaker (“We must continue to maintain social distancing.”), but it is comparatively infrequent (Table 8). These findings adhere to the preference for “should” over “must” in message 1 (self-efficacy) observed in the closed-questions. Other modal auxiliaries employed to express obligation are “shall” (“[…] we’re all in the same boat therefore shall we all follow the rules”) and “will” (“you will stick to the rules so […]”), though both are rare in this data.

Although the figures are too small to establish any statistically significant association, Table 8 shows that different modulation types involve different positionings of the speakers towards the obligations conveyed, explicitly construing the obligation as concerning them or others only. In contrast, all directives formulated in imperative mood were addressed to second person addressees.