2.1. Critical discourse analysis

CDA is an interdisciplinary approach that views discourse as both a product of societal influences, and a tool that actively shapes societal norms and power structures. (Fairclough, Reference Fairclough1995; Wodak, Reference Wodak1997); it understands texts as a product of society where power asymmetry can be articulated; that is why it contends that the latter can also be dismantled through detailed analysis (Blommaert & Bulcaen, Reference Blommaert and Bulcaen2000: 453; Wodak Reference Wodak1995: 204). Discourse analysts go beyond examining linguistic elements and surface-level descriptions to reveal the deeper meanings embedded in discourse, which is why this approach is referred to as Critical Discourse Analysis (CDA); it emphasizes the role of language in revealing underlying social, political, and ideological structures (Hidalgo-Tenorio, Reference Hidalgo-Tenorio2011: 187). Any corpus is perceived as a reflection of prevailing tendencies, as a mirror of hegemonic social practices, and, therefore, they should be analyzed within its context. In this vein, we will use it to explore power dynamics related to gender and ethnicity within certain segments of American society.

When corpora are studied, it becomes possible to conduct both corpus-driven CDA (Baker et al., Reference Baker, Gabrielatos and McEnery2013) and corpus-based CDA (McEnery et al., Reference McEnery, Xiao and Tono2006). Whilst the former uncovers linguistic patterns, and generates hypotheses directly from large, unbiased text corpora, corpus-based CDA tests predefined theoretical concepts using specific corpora, blending quantitative data with qualitative analysis to investigate particular aspects of language and its social implications. In this paper, both have been implemented. For our corpus-driven analysis, we use AntConc (Anthony, Reference Anthony2005) and Sketch Engine (Kilgariff et al., Reference Kilgarriff, Baisa, Bušta, Jakubíček, Kovář, Michelfeit and Suchomel2014); for our corpus-based analysis, we use UAM CorpusTool (O’Donnell, Reference O’Donnell2018). The use of both approaches represents a meaningful advancement in the study of language (Meyer, Reference Meyer and Meyer2014).

2.2. Sociolinguistics: Gender and ethnicity

Sociolinguistics studies the relationship between language and society (Johnstone, Reference Johnstone and Frawley2020: 124) from an interdisciplinary perspective (Bayley et al. Reference Bayley, Cameron and Lucas2013: 1). The many interests it has spurred in novel approaches, as well as the different goals and methodologies it has, has led to its development into several sub-disciplines: The variationist tradition (Labov, Reference Labov1966), ethnography of communication (Hymes, Reference Hymes1974), interactional sociolinguistics (Gumperz, Reference Gumperz and Dijk2005) and sociology of language (Fishman, Reference Fishman1971). Other approaches have originated in relation to topics such as ideology, gender, social class and ethnicity (Ricento, Reference Ricento, Bayley, Cameron and Lucas2013: 12–13). These extralinguistic variables have been studied alongside linguistic elements to discover how they coexist and interact.

2.3. Systemic functional linguistics: The Appraisal framework

Systemic Functional Linguistics understands language as a social semiotic system where meaning is made through choice. Despite the differences between Fawcett’s (Reference Fawcett2000) Cardiff School and Halliday’s (Reference Halliday1984) Sydney School (see Bartley, Reference Bartley2018b), both models support the existence of three linguistic metafunctions. The ideational metafunction is concerned with how individuals represent their experiences (Halliday & Matthiessen, Reference Halliday and Matthiessen2014: 186); here, the system of Transitivity is a key component. The systems of Modality and Appraisal are cornerstones of the interpersonal metafunction, which is related to how addressers and addressees interact (Halliday & Matthiessen, Reference Halliday and Matthiessen2014: 188), and evaluate their surroundings (Martin & White, Reference Martin and White2005; Halliday & Matthiessen, Reference Halliday and Matthiessen2014: 187). As for the textual metafunction, it deals with the way in which discourse is organized so that it achieves coherence through cohesion (Halliday & Matthiessen, Reference Halliday and Matthiessen2014: 191).

The Appraisal framework (Martin & White, Reference Martin and White2005) covers the interpersonal metafunction. It refers to how people make use of the semantic resources of a language to communicate, intake and mediate emotions, judgments and valuations (Wei et al., Reference Wei, Wherrity and Zhang2015: 237). This model explains how individuals assess their inner and outer realities through attitudinal shades of meaning coded lexico-grammatically. The Appraisal framework is integrated by three domains that are not symmetrical. Whilst Attitude is recognized as the central sub-system, Engagement and Graduation are considered peripheral (Wei et al., Reference Wei, Wherrity and Zhang2015; Li, Reference Li2016). Attitude is associated with evaluation in relation to emotion (Affect), ethics (Judgment), and aesthetic and social values (Appreciation) (Martin & White, Reference Martin and White2005: 42) Engagement is concerned with how the author’s voice is positioned in relation to their statements; the ways in which the interpersonal space and truth value of the text are negotiated (ibid.: 95). As for Graduation, it deals with degree; the meaning of utterances can be enhanced or diminished (Force) (e.g. I cried too many times), or the boundaries of entities can be sharpened (Focus) (e.g. He was a real father).

Although Martin and White’s (Reference Martin and White2005) framework can capture emotional evaluation in discourse efficiently, some of its categories were not as fully developed as others. Bednarek’s (Reference Bednarek2008) corpus-based refinement and Benítez-Castro and Hidalgo-Tenorio’s (Reference Benítez-Castro, Hidalgo-Tenorio, Mackenzie and Alba-Juez2019) amendment of the Affect sub-system try to address its areas for improvement. Consequently, in the present paper, we will adopt the latter approach (see Figure 1 below). The basic tenet of this development is the notion of goal. In theories of human evolution (see Darwin, Reference Darwin1872; Bazzanella, Reference Bazzanella and Bazzanella2004), emotion is defined as a mechanism for achieving survival-related goals; so, if emotions are goal-related, goals must be the central pillar of the Affect sub-system. In this new model, emotions are organized into three categories. Whilst goal-seeking emotions are cognitive and involve the surprise, interest and inclination triggered by any entity, goal-achievement emotions relate to whether emoters can or cannot attain or keep their goals; lastly, goal-relation emotions concern the attraction or repulsion individuals feel towards a specific trigger.

Figure 1. The Attitude system (Martin & White, Reference Martin and White2005; Benítez-Castro & Hidalgo-Tenorio, Reference Benítez-Castro, Hidalgo-Tenorio, Mackenzie and Alba-Juez2019).

As for the other two Attitude sub-systems, Judgment entails the author’s opinion about human behavior in two ways (Martin & White, Reference Martin and White2005: 53); on the one hand, social sanction has to do with how un/acceptable certain actions are within a specific society; this includes propriety and veracity. On the other hand, social esteem is linked to human conditions socially dis/approved of such as normality, capacity and tenacity. Appreciation is divided into three main sub-categories: Reaction is about the extent to which something or someone can grab the speaker’s attention; the extent to which an entity is cohesive and congruent is observed through composition; valuation has to do with whether an entity is worthwhile or socially relevant (Martin & White, Reference Martin and White2005: 67).Footnote ¹

2.4. Forensic linguistics

In The Evans statements: A case for forensic linguistics, Svartvik (Reference Svartvik1968) explains that Forensic Linguistics (FL) has a three-fold nature, based on which its focus is, namely, the study of written discourse in law, the analysis of discourse during a trial, and the analysis of interaction throughout the whole legal process (Coulthard & Johnson, Reference Coulthard and Johnson2010). Moreover, as Ariani et al. (Reference Ariani, Sajedi and Sajedi2014) explain, FL is used in several areas such as author identification, forensic stylistics, discourse analysis, dialectology, forensic phonetics, forensic transcription and variation. Considering the abovementioned, the present paper is a soft version of FL, where we carry out a discourse analysis of inter-author variation in a corpus of post-exoneration interviews.

5.1. RQ1: What emotional shades of meaning are conveyed in the narratives of these wrongfully convicted individuals?

Researchers suggest that individuals who have been wrongfully convicted often experience victimization and social stigma, leading to lasting psychological impacts (see Ricciardelli et al., Reference Ricciardelli, Bell and Clow2009; Clow & Leach, Reference Clow and Leach2015). Supporting evidence is presented in the keyness analysis in Table 2, which identifies a variety of terms associated with their imprisonment and the broader legal process (e.g. death date, death row, court date, new trial), and references to institutional agents involved in this traumatic scenario (e.g. arresting officer, state trooper, lay guard).

Table 2. List of KW of the whole corpus

There seems to be a stronger focus on the consequences of the crime than the incident itself. This makes us think that the subjects may have been able to internalize the victims’ accidental death but are unable to overcome the trauma of their own unjust imprisonment. This idea is further reinforced by the institutional agents’ being indirectly addressed without mentioning their names; William Sizemore, the actual murderer in one of the crimes, is named explicitly, though. This dichotomy underscores a complex psychological response. The subject remembers the criminal’s name but is unable to respond appropriately to their aggressor. This may suggest dissociative mechanisms, where memory and behavioral responses are disconnected due to trauma, indicating a deep struggle in processing such events. Victims may verbalize details yet remain ineffective in confronting the abuser, possibly as a protective coping strategy. Additionally, the absence of specific names for institutional agents may enable victims to articulate instances of institutional maltreatment more freely. Crucially, contextual non-sentient elements have to do with dates, as in court date, and also with destruction (e.g. death date, death row, same exact disease). Thus, the notably high frequency of these terms uncovers negative attitudinal meanings associated with the experience itself. From all these time-related expressions, we may infer some anxiety. Moreover, not all negative emotions are event-based, since some of them come from other individuals or the subjects under analysis. For example, phrases like made a mistake and mistake in cases may denote a prevailing sense of dissatisfaction, or even guilt, among those involved. Generally, the discourse is characterized by destructive events that adversely affect the mental well-being of the speakers; this reflects how linguistic choices can underscore the emotional and psychological toll of their experiences, indicating a broader impact on their state of mind. Nonetheless, there is a positive element in the whole list. The Innocence Project, the organization that granted their exoneration, is the only item in which dissatisfaction is left aside. Furthermore, it is extremely relevant to see how metaphors can be used to describe space, for example, the cell in which the police kept them: “It was just stayed in this bullpen like you do dogs with no shade.” (FE_AF_SB). In this case, the subject uses this discursive strategy to show that they were treated like animals; this comparison is intended to highlight the dehumanization they experienced, emphasizing how they were reduced to something less than human in the eyes of others.

Figure 2 explores the overall evaluative load of the corpus. The results of Valence (emotion) and Axiology (opinion) agree with the aforementioned. At first sight, we can see that cases with unpleasant emotion and negative opinion are twice as frequent as those with positive emotions and opinions.Footnote ³

Figure 2. Attitudinal evaluative load in the corpus.

In most cases, pleasant emotions are related to experiences prior to the crime. In Example 1 below, calm refers to an event that is not directly connected to the death of the subject’s sons but rather to the emotional bond with Larry’s father. Additionally, the few cases in which there is some evidence of positive emotion after imprisonment, these are triggered by external events. This is the case of Example 2, where Kristine implicates how she was positively surprised when discovering in jail that she was pregnant. Positive surprise is inferred here because she later goes on to say that her son Trent was “the miracle that saved me” (FE_CA_KB), which also denotes positive axiology.

1. (1) I felt calm because my father had taken me everywhere with him when I was younger and older in the station wagon. (MA_CA_LD)

2. (2) And in the midst of that, I started getting sick, and that’s how I discovered I was pregnant. (FE_CA_KB)

Similarly, positive opinion is also limited to the subjects themselves (Ex. 3), and sometimes to family members (Ex. 4) or to the victims (Ex. 5).

1. (3) And I have so many opportunities. (FE_CA_KB)

2. (4) She gave me the car. (MA_CA_LD)

3. (5) Why would an innocent baby be taken from us? (FE_AF_SB)

We can see, then, that the exonerees can express their opinion but avoid appraising institutional agents. As it can be expected from the context and content of the interviews, unpleasant emotions and negative axiology are omnipresent. Furthermore, its pleasant and positive counterparts are distant from the event, since they are related to previous and latter experiences as well as other subjects.

5.2. RQ2: How does gender influence their discourse?

Departing from the definition of genderlect (Tannen, Reference Tannen1990) as a linguistic variety whose specific traits are solely related to the male–female dichotomy, the following discussion will challenge whether any linguistic differences exist and whether said differences are the ones traditionally expected for each gender. Before going further, it is relevant to mention that the results come from a fine-grained annotation scheme; therefore, deviations are expected.

Turning now to the Appraisal-based analysis, all the findings will be presented with their raw data, the percentage calculated by UAM Corpus Tool (O’Donnell, Reference O’Donnell2018), as well as their Chi-squared score.

As illustrated in Table 3, Explicitness and Implicitness are two primary categories that significantly influence the expression of emotion and opinion. Explicit emotions and opinions differ between the two groups. Whilst the males tend to showcase inscribed emotion, the females opt for explicit opinion. Interestingly, both men do not just say how they feel explicitly, they also report other people’s feelings in a direct way.

1. (6) And I got angry at him for even suggesting something that heinous was done to any child. (MA_CA_LD)

2. (7) He loved station wagons. (MA_CA_LD)

   Table 3. Explicitness and Implicitness in each gender group

   

The fact that Larry DeLisle and Ru-el Sailor are able to overtly articulate their inner worlds, their emotional experiences, especially regarding the crimes, suggest that they may have transcended their traumatic wounds. This observation challenges conventional assertions about the emotive nature of the so-called ‘female language’. Additionally, it is noteworthy that both women exhibit a greater inclination towards explicit expression of opinion, while both men tend to favor implicit forms of opinion. Sabrina Butler and Kristine Bunch discuss how external institutional forces (also addressed as they) were abusive towards them without reservation, and denounce and condemn their harmful behavior (Ex. 8). In contrast, men do not make explicit judgments about institutional actors; they merely implicitly hint at a fraudulent reason behind their actions (Ex. 9). The latter can be explained by taking a closer look at the subcategories for the Achievement sub-system.

1. (8) […] they really, really, really do not like to admit they made mistakes in cases where they contribute to an 18-year-old young woman being sentenced to death for something she did not do. (FE_AF_SB)

2. (9) He refused to do it because the arresting officer probably did not want them to find it. (MA_CA_LD)

In Table 4, we can see that, whereas satisfaction is constant for both males and females due to the nature of the situation, women feel a significantly higher degree of dissatisfaction than men. Significantly, this dichotomy can be ultimately linked to a generalized feeling of fear in the female group (see Figure 3).

Table 4. Achievement types in each gender group

Figure 3. Fear in each gender group.

Police misconduct and social stigmatization seem different among the two genders. Women in this corpus reflect this sense of insecurity explicitly and implicitly. This could be a side-effect of the nature of the events, since both Sabrina and Kristine did want to save their children but were unable to do so due to external forces. Thus, the fact that they can pinpoint the (negative) tenacity of the police officers and the negative propriety of the people witnessing the incident is essential for understanding their position.

1. (10) They kept telling me, no, you, you cannot go in and know we can get him and know he is alive. (FE_CA_KB)

2. (11) The first lady that opened her door, she said that her kids were sick and she did not have time to take me to the hospital. So that lady closes the door in my face. (FE_AF_SB)

Sabrina and Kristine feel dissatisfaction and fear in relation to the whole event. From the start, despite their positive tenacity and efforts to save their children’s lives, they find themselves in a state of helplessnes. Whilst Kristine is kept from getting into the fire to grab her son, Sabrina is neglected and feels useless. Both connotative and denotative fear permeate the whole narration. Police maltreatment and general improper treatment make the two women even more discontent and scared. Alongside everything that had happened, when in prison, Kristine, who got to know that she was pregnant, was taken to a hospital where her baby’s heartbeat could not be heard (Ex. 12). Meanwhile, Sabrina was mistreated during police interrogation and her imprisonment, which ultimately led fear to pervade her depiction of the whole experience (Ex. 13).

1. (12) I did not really know that I was in labor because I was just having a backache, so they took me to the infirmary and could not find a heartbeat. (FE_CA_KB)

2. (13) And after so many hours of him yelling and screaming and looking like he wanted to get up and jump on me, fight me, you know, I was scared. (FE_AF_SB)

Nonetheless, despite feeling more fear of institutional entities, this deviation is not statistically significant since both males and females showcase similar rates for the appraised sub-category.

In relation to the appraised types in each gender group, Table 5 shows a divergence as to the overall authorial versus non-authorial dichotomy. In this case, the two women are evaluated more frequently than men (by themselves and others) in terms of veracity, tenacity and capacity, which may explain their instilled dissatisfaction. The fact that they say they were telling the truth, being tenacious and capable may be interpreted as a way to counteract the agents’ negative acts. Interestingly, whilst females discuss their veracity (Ex. 14), capacity (Ex. 16) and positive tenacity (Ex. 18), they use the same categories for negatively judging institutional agents; in other words, these oppressive entities are depicted as deceitful (Ex. 15), ineffectual (Ex. 17) and displaying stubborn resistance (Ex. 19).

1. (14) I finally started telling him what exactly had happened. (FE_AF_SB)

2. (15) He wrote on the statement that I had punched my son. (FE_AF_SB)

3. (16) I can make it through anything. (FE_CA_KB)

4. (17) The police did not help because they followed me everywhere I went (FE_CA_KB)

5. (18) I was trying to get him to breathe. (FE_AF_SB)

6. (19) They were dead set on saying you stomped your baby, beated them. (FE_AF_SB)

   Table 5. Appraised types in each gender group

   

In the case of men, non-sentient entities were expected to be more outstanding, since both of them emphasize the material instrument used in the crime (that is, a gun in the case of Ru-el, and a car in the case of Larry). This tendency is reflected by the fact that they evaluate the crime itself and its repercussions through indirect appraisal. Object appreciation is especially central in relation to maintenance and utility; in fact, crime instruments are depicted as destructive (Ex. 20) and useless (Ex. 21). In categorizing the objects this way, Ru-el and Larry take the guilt from them and place them on the inanimate entities involved, which can be translated into the implicit defense of their innocence (see Table 6).

Table 6. Judgment and appreciation types in each gender group

1. (20) We had a certified AC master mechanic who found all the problems with the vehicle. (MA_CA_LD)

2. (21) When the accelerator stuck, it threw me off. (MA_CA_LD)

Additionally, in relation to the appraised entity, we must also mention that Sabrina and Kristine discuss the event of their exoneration a significantly higher number of times. They talk about their release day in a bittersweet way. Although they feel satisfaction for being free again and re-encountering with their other children, at the same time, they have a deep sense of despair after they lost one, and fear that they cannot successfully fit in society even after exoneration.

1. (22) After that initial ‘oh’ wears off, then you are left with all of these fears and insecurities. (FE_CA_KB)

2. (23) I wasn’t and I’m still not free. (FE_AF_SB)

The latter shows that wrongfully convicted individuals can be subject to long-life victimization (Ricciardelli et al., Reference Ricciardelli, Bell and Clow2009; Clow & Leach, Reference Clow and Leach2015). They carry an invisible wound that scars them for a lifetime. The sense of full freedom is taken from them, making them feel detached from society and unable to directly appraise the perpetrators of their pain (see Hidalgo-Tenorio & Benítez-Castro, Reference Hidalgo-Tenorio and Benítez-Castro2021).

5.3. RQ3: How does ethnicity influence their discourse?

Let us consider the last variable for the present paper. As discussed in Section 2.2.2 , ethnicity is a central factor in sociolinguistics that can reveal asymmetrical power relations (Heller & McElhinny, Reference Heller and McElhinny2017: 8).

Modality is the first category where contrast emerges. The system of modality was initially incorporated into the annotation scheme to explore the dimension of interactivity (Bartley & Hidalgo-Tenorio, Reference Bartley and Hidalgo-Tenorio2016: 2–3). This includes subcategories such as probability, possibility, certainty, obligation, advice, permission, volition, prohibition, ability and evidentiality. As reflected in Table 7, only possibility, certainty and ability show some interesting differences: While Caucasian subjects demonstrate significantly higher rates in relation to possibility and ability, the African-American group exhibits a weaker significance in terms of certainty.

Table 7. Modality types in each gender group

Possibility is the first category in which the two groups showcase statistically significant differences. Caucasian individuals display high rates of its general use and tend to opt for modal verbs for its realization. By doing so, they explore alternatives in relation to the events mentioned, and implicitly portray themselves in a positive light.

1. (24) It would prove what I was saying from the beginning. (MA_CA_LD)

2. (25) I can make it through anything. (FE_CA_KB)

Despite this, the negative sub-category remains consistent across both groups, indicating that African-American individuals do not necessarily view events as impossible; rather, they tend to avoid perceiving and labeling things as possible. This may be linked to their perception of a subordinate position in society (Lo, Reference Lo2020: 77), a side-effect of the historical oppression these have suffered in the United States (Rosa & Flores, Reference Rosa and Flores2017: 645). In contrast, Caucasian individuals perceive events as possible to a higher degree.

Similarly, the category of ability can be understood through the same ingrained perspective. Caucasian individuals display higher rates of material ability, particularly concerning the incident. However, in contrast to the previous category, this group incorporates both the ‘able’ and ‘unable’ sub-categories in their speech.

1. (26) I got there as quickly as I could. (MA_CA_LD)

Lastly, regarding certainty, the rates are reversed, with African-Americans displaying somewhat greater certainty than Caucasians. This may indicate that they are confident in their statements, but it could also reflect a need to employ confirmatory devices when expressing their opinions. For instance, in Example 27, Sabrina explicitly states that she was fully aware of their intentions toward her.

1. (27) I just knew that they were going to come back and say I was guilty. (FE_AF_SB)

Moving on to the axiology-valence dichotomy, African-American and Caucasian subjects display similar rates in expressing opinion, whereas their portrayal of emotion does vary (see Table 8). Additionally, in Figure 4 we can observe that the Caucasian group exhibits higher rates of overall valence, including both pleasant and unpleasant emotions.

Table 8. Valence and axiology in each ethnicity group

Figure 4. Valence types in each ethnicity group.

The rates for Explicitness are also significantly higher concerning valence (see Figure 5); in other words, Caucasians openly discuss both the positive and negative emotions related to the context of the crime, and, consequently, their unjust incarceration, whereas African-Americans in this particular study tend to refrain from doing so (Ex. 28). Despite this fact, no significant differences are found in relation to the implicit account of emotion. So, it is not that they express their feelings in a particular manner; rather, they generally avoid expressing them altogether.

1. (28) […] it just makes me angry that they are going to point to the fact that I’m not classical as far as how I was supposed to react to losing my children. (MA_CA_LD)

   

   Figure 5. Explicit emotion by ethnicity.

The latter is especially relevant because it relates to previous findings in relation to trauma. In Hidalgo-Tenorio and Benítez-Castro (Reference Hidalgo-Tenorio and Benítez-Castro2021), we find that abused women often silence their emotions and choose not to mention the agents of their suffering; similarly, we can observe comparable patterns in cases of miscarriage of justice. Within this context, African-American people could have been victims of police brutality since they neither show emotion in any way nor mention their abusers’ names; they instead distance the action from themselves without explicitly referring to their discomfort.

1. (29) There was a vice cop from my neighborhood that particularly did not get along with me. (MA_AF_RS)

Furthermore, African-American individuals showcase higher rates of neutral emotion than Caucasian people do:

1. (30) I just wanted it all to end. (FE_AF_SB)

The example above illustrates how African-American subjects use lexical modality (volition) in order to neutrally express something that otherwise would be considered as negative valence. In this case, the desiderative mental process of wanting something to end denotes an underlying dissatisfaction; the context of the sentence gives us information about the hidden (negative) emotive load of the sentence (i.e. “You know, they kept screaming at me and that was like four hours of interrogation with him”, FE_AF_SB). Once again, abused people seem to avoid opening up; thus, negative emotion is softened and subjects are silenced.

In the case of axiology, the two groups are radically different in relation to some of its sub-categories. This is true for the appraiser, appraised and judgment types. Firstly, although both show similar rates for the authorial appraiser sub-category, the African-American group reflects a higher number for the non-authorial appraiser sub-category; that is, the first-person narrator reports someone else’s opinion from a second- or third-person perspective (White, Reference White1998): “He kept saying, you know, this is what you did” (FE_AF_SB).

By doing so, the author is not emotionally involved (Ananda et al., Reference Ananda, Nababan and Santosa2018). Interestingly, the appraised category shows a mirror image of the previous tendency. Although they do not voice their own views, African-Americans are the center of evaluation: They are evaluated by themselves (e.g.“I was a poor teenager, that didn’t have a job or money”, FE_AF_SB) and by other external voices (e.g. “He had a good informant, told him that I was the driver, that I was the shooter”, MA_AF_RS). Although at different rates, Caucasian individuals are also subject to evaluation. Nonetheless, most judgments and appreciations come from themselves, and they all have some kind of emotive tone.

1. (31) I’m very good at what I do. (MA_CA_LD)

2. (32) I have played it over and over and over and over again. (FE_CA_KB)

Lastly, both groups differ for overall judgment, as well as three of its sub-categories, namely, propriety, veracity and tenacity.

Table 9 above reflects a higher tendency for judgment in African-American subjects; that is, they show a strong preference for evaluating human behavior in relation to ethical values (Hidalgo-Tenorio & Benítez-Castro, Reference Hidalgo-Tenorio and Benítez-Castro2021: 317). Once again, this fact does not mean that they are explicitly assessing entities and events, since, as it has been mentioned earlier, evaluation in this group is widely done through a non-authorial appraiser. Additionally, this table shows all three sub-categories as well as their polar opposites; nonetheless, not all of them could be considered as such, since often its degree of negativity depends on the trigger (Benítez-Castro & Hidalgo-Tenorio, Reference Benítez-Castro and Hidalgo-Tenorio2022). Propriety reflects higher rates for African-Americans, especially in relation to unethical behavior (e.g. “The lead investigator kept screaming at me”, FE_AF_SB). This fact also reinforces the hypothesis that they are both subject to social discrimination and police abuse, which eventually leads to a silencing of their traumatic experiences. Similarly, the veracity category is mostly used by this group (e.g. “He played with my innocence to get free, and now he told the truth”, MA_AF_RS). Lastly, the numbers for the tenacity sub-category in the African-American group are three times higher than those in the Caucasian group. As aforementioned, determination is not always intrinsically positive since it depends on its context. Bearing this in mind, the fact that the tenacious class is significantly higher than its weak counterpart does not imply a positive appraisal. To explore this fact, we will take a closer look at some examples:

1. (33) I’m going to keep fighting to keep evidence as I go. (MA_AF_RS)

2. (34) So, I did not sign where he told me to sign. (FE_AF_SB)

3. (35) He kept saying, you know, let me take a look at this case. (FE_AF_SB)

4. (36) They followed me everywhere I went, and talked to everybody. They even showed up at my son’s funeral. (FE_CA_KB)

Table 9. Judgment types in each ethnicity group

Example 33 showcases a strong determination from the part of the speaker to get evidence after getting in contact with the Innocence Project. Here the category tenacity would coincide with the view of the former as a positive characteristic. Similarly, Example 34 reflects a positive tenacity on the part of the subject, which happens when a police officer abuses her into signing a false declaration and she refuses to sign where she was forced to. This small act of defiance would have been appraised differently had the speaker not been herself. In contrast, Example 35 presents a positive perspective on a non-authorial entity. Sabrina states that Clive, her second attorney, was really interested in helping her, which he eventually did. Lastly, Example 36 reflects a negative appraisal of the police officers, which is due to their strong determination to condemn her; from Kristine’s perspective, they are not just doing their job but are extremely perseverant in putting pressure on her.

Just to finish this section, we will briefly comment on transitivity. As mentioned above (see Section 2.3), the system of transitivity is a key element of the ideational metafunction (Halliday & Matthiessen, Reference Halliday and Matthiessen2014: 186); it serves to identify how subjects construe their inner and outer worlds. In our corpus, significant differences have been found regarding verbal processes, which are associated with the notion of symbolic exchange of meaning (Halliday & Matthiessen, Reference Halliday and Matthiessen2014: 303). Particularly, as Figure 6 shows, the African-American group triplicates the numbers of the Caucasian group. Curiously, they do not use verbal processes to voice their own stories (see Ex. 37); once again, they voice their opinions through other people’s verbal actions (see Ex. 38). All in all, there are also examples in which we can hear them directly speak directly for themselves (see Ex. 39).

1. (37) None of the witnesses in this case described the two individuals as two light skinned males that looked like brothers. (MA_AF_RS).

2. (38) She did not tell me the right way. (FE_AF_SB)

3. (39) I tell my son Trent all the time that he’s the miracle that saved me. (FE_CA_KB)

Figure 6. Verbal processes contrasted by ethnicity.