Agency, Responsibility, and Actor Positioning In Courtroom Narratives

Viewing language as a system consisting of grammatical resources for meaning making, this study explores how agency and responsibility are attributed in legal narratives through the lens of transitivity. Drawing upon the opening address of three American trials, the quantitative and qualitative findings indicate that agency and blameworthiness of the individuals on trial are discursively negotiated through starkly different grammatical choices, so that polarized positionings of the same social actors and events are accomplished for the audience. It is argued that such manipulation of grammatical resources exhibits subjective intervention on the part of the presenter and constitutes a prime mechanism of inference and attitudinal evocation for the jurors. In effect, the opening statement, which is in principle intended to be merely informative, becomes not only argumentative but also evaluative.


Introduction
sequences in display talk, prevalent in other phases of a trial. Known for its primacy power, the opening speech allows the jury to vividly remember the information that is presented first, thereby leading them to draw tentative conclusions early on (Lind and Ke 1985;Pennington and Hastie 1991;Schuetz and Lilley 1999;Spiecker and Worthington 2003). There has also been some empirical evidence that confirms this primacy effect. Pyszczynski et al. (1981), for instance, find that jurors claim that they recall hearing specific testimony in the evidence phase, when in fact such testimony is only promised in the opening event but is never fulfilled later on. By law, an opening statement cannot be argumentative; that is, it cannot challenge the truthfulness or credibility of the witness or the opposing party. The United States Supreme Court has limited opening statements to apprise the trier of fact of the issues in question and to summarize the evidence that each side intends to offer during the trial (Snedaker 1991). That is, they are "intended to do no more than to inform the jury in a general way of the nature of the action and defense so that they may better be prepared to understand the evidence" (Best v. District of Columbia, 291 U.S. 411, 54 S. Ct. 487, 78 L. Ed. 882 [1934]); my own emphasis). What this means in practice is that lawyers can, for example, offer a discussion of the anticipated evidence and "facts" they intend to prove, but they cannot assert personal opinions, comment about evidence, discuss how to apply the law to the facts, or arouse the emotions of the jurors by making positive or negative judgments about the defendant. In reality, the boundaries between informative and argumentative content cannot be straightforwardly enforced and are usually left to the discretion of the trial judge.
Indeed, past research has revealed that opening statements are essentially argumentative. Cotterill (2003) examines the role of "strategic lexical choices" in framing courtroom narratives. To illustrate, the prosecution's choice of "encounter" and "control" embodies negative semantic prosodies, thereby conceptualizing O.J. Simpson as a violent man, while the defense diffuses such an image by lexicalizing talk in the Simpson household as "discussion", "conversation" or "dispute", all of which are verbal, rather than physical, thereby countering the physical violence and capacity for murder. Heffer (2005, 77) also finds evidence of authorial intrusion and explicit comments throughout the narrative, such as "The prosecution would say that is an obvious piece of dangerous driving". Chaemsaithong (2018) shows how reference terms ascribe polarized identities for both the defendant and victims in each side's narrative. For example, the prosecution team almost exclusively refers to the person on trial as "defendant", a term that solely positions the referent according to his legal status and suppresses other identities outside the courtroom (such as being a father or being a good friend of someone). In contrast, the defense team refers to the same individual by his first name, thereby individualizing and creating solidarity with the person on trial. When it comes to the victims, the prosecution refers to them by nickname and age, while the defense avoids making any reference to them. As a result, issues about damage and injuries are foregrounded and backgrounded accordingly. All in all, previous research shows how different linguistic features transform opening speeches into argumentative texts.
The current study takes as its starting point the argumentative nature of the opening statement and goes further to explore ideological positioning in this courtroom genre. Unlike previous studies, the focus lies not in lexical choices but in grammatical choices that lawyers make. Underpinned by the functionalist view that language is understood as "networks of interlocking options" (Halliday 1994, xiv), the study seeks to explore how agency and responsibility are attributed to the person on trial by quantitatively and qualitatively scrutinizing clausal configurations through the lens of transitivity. In this article, agency is understood as the property of those entities that possess some degree of control of their behavior, whose actions in the world affect other entities' (and sometimes their own), and whose actions are the object of evaluation in terms of their responsibility for a given outcome (Duranti 2004, 453). As such, agency has a strong link with responsibility, and causality (Dreyfus 2017). Agents who act voluntarily, by their own choice, are held responsible for their actions, which are usually directed towards and cause an effect on another agent. Those agents acting under causal constraint, on the other hand, are not to be so responsible and do not impact another agent (Hopper and Thompson 1980, 254;Bakker 1994, 25;Barnes 2000, 5). From a discourse analytic perspective, agency, responsibility and causality are constructed and negotiated in interaction, and can be used to position a social actor as powerful or powerless or as acting with or without his free will (Ostman and Solin 2016), through different linguistic devices, such as reported speech (Johansen 2011), activepassive constructions (Pomerantz 1978;Henley, Miller, and Beazley 1995), causal constructions such as "because" or "as", which can convey or deflect responsibility (Brown and Rubin 2005), and verb choices (LaFrance and Hahn 1994;O'Connor 2000), to name but a few.
Based on three high-profile criminal trials, the analysis endeavors to answer the following questions: 1) How are the issues of agency, responsibility and causality mediated through specific grammatical choices in opposing lawyers' narratives? 2) How do the two sides differ in that meaning making process, and 3) How are such differences indexical of the prosecution's and the defense's ideological stances? As we will see, the interpretations of guilt and innocence of the social actors are characteristically encoded in specific grammatical choices that each side makes, so that the same people and events emerge as two completely opposite representations. It is argued that such manipulation of grammatical resources conceals subjective intervention on the part of the presenter and constitutes mechanisms of inference and attitudinal association for the jurors.
This article starts with a detailed discussion of agency and how it can be transmitted in the grammatical system, and proceeds to give an overview of the corpus and methodology. It then presents the findings in detail and concludes with some observations on the common grammatical patterns found in this genre.

Agency, Responsibility and Grammar
In Halliday's functional grammar framework (1994), agency is grammatically encoded in language in the system called "transitivity", which refers to how a speaker ideationally encodes his experience of the external world and the internal world of his own consciousness, together with the participants and attendant circumstances, and organizes those pieces of information into a clause. Consisting of structures for the representation of speakers' experience, transitivity allows language users to express and establish relations as to "who does what in relation to whom/what, where, when, how and why" (Hasan 1988, 63). Choices with respect to which participants are represented as agentive and which as affected often have a potential to influence who is to be seen as more to blame and who as less to blame (Trew 1979;White 2006). The degree of blameworthiness can be conditioned by the degree to which they are represented as playing an active, agentive role. Trew (1979), for instance, shows how two British newspapers implicitly position the reader to blame the so-called "rioters" by presenting them as "agent" in a high proportion of the clauses (as in "Rioting blacks shot dead by police."), while another newspaper positions the reader to view the police as more to blame by presenting them as the primary agents and the crowd as acted upon (as in "Police shot 11 dead in Salisbury riot."). Trew (1979, 99) also finds that the passive voice pattern, as in "Eleven African were shot dead…when Rhodesian police opened fire on a rioting crowd", can "shift the focus away from those who did the shooting and onto the victims", due to the fact that agency for the shooting has to be inferred from the context. By attributing violence to black people, the researcher argues that this reflects a more widespread ideological perspective used by the dominant white elite in Rhodesia to justify white rule in Africa. Suffice it to say, therefore, that "linguistic codes do not reflect reality neutrally; they interpret, organize, and classify the subjects of discourse. They embody theories of how the world is arranged: world-views or ideology" (Fowler 1986, 27). In courtroom discourse, ascriptions of agency and responsibility not only constitute a central concern for attorneys and judges but also play a decisive role in the outcome of the trial (Ehrlich 2001;Dupret 2003;D'hondt 2009). To prove the defendant's criminal liability, the prosecutor must somehow demonstrate that the defendant has committed an overt criminal act with a criminal intent. For this reason, descriptions of the case that seek to inculpate the defendant must allow the jury to infer that the individual on trial has acted intentionally, caused harm to others, and possessed the capacity to exercise control over his actions, while the defense works to contest such descriptions. Komter (1998, 49-50) identifies a defensive strategy that she characterizes as "the disappearance of agency", evidenced in the following example: "Then I suddenly remembered my gun and I reached for it and then there were shots". Here, the suspect describes a shooting that he allegedly committed as if the shots just seem to have been there, irrespective of his actions. Examining rape cases, Ehrlich (2001) focuses on what she calls "the grammar of non-agency", that is, the strategy that the defendant and his representatives use to diffuse, obscure or eliminate the defendant's agency, instead of denying outright the occurrence of consensual sexual activity. For instance, when asked by the lawyer whether "[y]ou're the one who started to kiss her [the complainant] first", the defendant answers, "I lied next to her and we laid there for about two minutes and then we both came and kissed each other mutually" (Ehrlich 2001, 46). The reciprocal verbs "came and kissed" conflate both the defendant and complainant as the agent and recipient at the same time, thereby portraying the alleged rape as a mutual endeavor. Empirical support from the psycholinguistic literature also exists for the interpretations particular linguistic forms can create. Studies by LaFrance and Hahn (1994), and Henley, Miller, and Beazley (1995) find that changes between active and passive voice affect the interpretation of statements; specially, the subjects tend to attribute greater causality to patients, as opposed to agents, when interpreting sentences in the passive voice.
According to Halliday (1994), two major key components in the transitivity system include the process and the participant. The former refers to how an event is expressed through verb phrases. An event may be represented as actions (material process); speaking events (verbal processes); states of mind (mental processes); states of being or quality, possession, or location (relational chaemsaithong MANUSYA: Journal of Humanities 24 (2021) 204-226 processes); physiological and psychological behaviors (behavioral processes); or existence (existential processes), as exemplified in Table 1.
Choices of processes are key to meaning making for at least three reasons. First, lawyers may choose different processes to represent the same event, each expressing different degrees of agency and blameworthiness. To illustrate with examples that I create, the prosecution may recount an event using a material process, as in the defendant killed the victim, while the defense may choose an existential process for the same event, as in there was a killing, or may modify it in some ways, as in the defendant did not kill the victim. Second, the same process may be consistently repeated (or avoided), and so certain aspects can be highlighted or concealed, as in when the lawyer uses the same verbs many times in his narratives for emphasis. Third, processes which are of the same type but which express different degrees of evaluation may be chosen. For instance, the verbal process of the defendant conceded he committed the crime implies that the defendant must have attempted at some point to deny committing the crime, or at least had the intention of so doing, while the verbal process of the defendant admitted that he committed the crime portrays the defendant as more honest. What these examples show is that events can be accurately reported but, at the same time, the ideological mark of the presenter can be worked into the story.
Occurring alongside processes are participants, realized by noun phrases. In the Hallidayian literature, these participants are given different labels, depending on the process type in which they occur (e.g. "actor" and "goal" table 1 Examples of process types (selected from the data under study)

Process types Examples
Material Conrad Murray gave propofol to Michael Jackson./ Brown was going to school with his aunt.

Verbal
The defendant talked about shooting that man. Mental Jackson needed help. Relational -quality -possession -location He was a landscaper. The defendant had no savings accounts./ The witness will corroborate this defendant's statement to the police: "We did recons...It is planned." Jackson wasn't at the rehearsals. Behavioral Michael Jackson died.

Existential
There was a robbery in Baton Rouge. in material processes; "carrier" and "attribute" in relational processes, etc). However, broadly speaking, participants are either the doers of the process, ones who have agency and, in many cases, power to affect other people, or the done-to's, who have no agency and are at the affected, receiving end of action (Fairclough 2000). Interestingly, participants may be made optional through the grammatical operation called "nominalization", in which the process is transformed into a noun phase. This can also be ideologically important (Fairclough 2000, 26), because the operation removes responsibility, consequences, and other specific details, and makes these issues a secondary concern. Aldridge and Luchjenbroers (2007) reveal that in a sexual assault case, the defense lawyer minimizes the violence and abuse that the victim experiences by referring to the event as "a sexual encounter" rather than spelling out the actual processes. Similarly, Rosulek (2015, 85) finds that the defense uses nominalization to downplay the defendant's agency and the dynamic nature of events, as in "the incident of the killing of the animals took place" and "in response to Mr. Wilder's acts of what he did".
Despite its wide applications in various kinds of texts, the framework of transitivity has not been put to much use in courtroom narratives. One of the few studies is Figueiredo (1998), which examines transitivity choices made by five appellants in rape cases. During the legal proceedings, material processes occur more frequently than other processes, and this is likely because evidence presentation depends on justifying the events of the crime. The victims also use verbal or relational processes more often (for instance, "The girl, who was a virgin, did not consent to any sexual activity") (Figueiredo 1998, 106), so that their agency is deemphasized, and as a result the reader can sense their helplessness. Meanwhile, the male attacker is portrayed as the actor and consequently viewed as more active and more powerful.
Analyzing witness examination, Guang (2010) analyzes eight Chinese court cases, and finds material processes occur most frequently, followed by verbal processes, relational and mental processes, respectively. Behavioral and existential processes are scarce. While the afore-mentioned tendency may be found in everyday narratives because "doing" is the most common process in driving the story forward, what is interesting is that the defendant appears as the agent in material processes more frequently in the prosecutor's discourse than in the defense's. This is attributable to the prosecutor's goal to portray the defendant as active and aggressive. What is missing is the roles of victims in the clauses.
Examining closing statements, Rosulek (2008;2015) finds that the prosecution foregrounds the defendants and their agentive roles in the crime more chaemsaithong MANUSYA: Journal of Humanities 24 (2021) 204-226 than the defense does. This is because the defendant wants to de-emphasize the defendants' roles in the events. The victims also appear in the prosecution's discourse more frequently, while the defense remains silent and includes different and less information about them. While all of the studies above have done much to inform the present study, it remains to be seen whether the tendencies in these genres will hold in the opening statement.

3
Data and Analytical Process 3.1 Data A background description for each trial under study is given below.
Case 1: Commonwealth of Virginia v. Lee Boyd Malvo (Prosecution: 762 clauses; Defense: 1,125 clauses) The case features a 17-year-old defendant Lee Boyd Malvo who stood trial for his role in the 2002 sniper attacks around Washington, D.C. He, together with 42-year-old John Muhammad, was allegedly involved in killing ten people and wounded three others. However, it remains unclear who actually pulled the trigger in each of the shootings. The prosecutor argued that Malvo was guilty based on the fingerprint analysis of the rifle used and Malvo's confession to committing many of the shootings during the interrogation. The defense claimed that Malvo was brainwashed by Muhammad into helping to commit these criminal acts and was trained and de-sensitized with computer games so that he could shoot humans. Malvo was found guilty and sentenced to life in prison without parole (while Muhammad faced a death sentence in a separate trial).#8232; Case 2: The State of Texas v Darlie Lynn Routier (Prosecution: 304 clauses; Defense: 349 clauses) This case involves an American mother from Texas, Darlie Routier, who was accused of murdering her 5-year-old son Damon. The defendant made a call to the police to report that her home had been broken into and that an intruder had attacked her and stabbed two of her sons, Damon and Devon, who were asleep downstairs with her. The two children sustained fatal injuries and died, whereas Routier's wounds were superficial. The prosecution theorized that the defendant murdered her sons because of the family's financial difficulties and staged the crime scene before calling the police. The defense argued that there was no reason for her to kill her children and that the case did not have a motive, a confession or any witnesses. Routier was convicted of murdering only Damon (as the prosecution decided to ensure the option to Conrad Murray, a physician, was implicated in the death of world-famous singer Michael Jackson, due to a lethal dose of anesthetic propofol. During the trial, the prosecution recounted how Murray abandoned his practice to work exclusively for Jackson, how he ordered gallons of propofol in response to Jackson's complaints about insomnia, and how Murray administered propofol on the morning of Jackson's death. The defense did not dispute that Murray was responsible for propofol being in Jackson's home, or that Murray left Jackson alone after administering a dose of it on the morning of his death, but suggested that it was Jackson himself who administered that last injection. Murray was ultimately found guilty of getting Jackson close enough to propofol and giving him the opportunity to take it unsupervised. The verdict was a fouryear prison sentence (although he was released 2 years earlier, due to a change in California law).

3.2
Analytical Procedure As transitivity is expressed through process choices, the basic unit of analysis for this project is the clause level, including main and subordinate clauses as well as finite and non-finite clauses. The first step involved identifying clauses in which the main social actors (i.e. the defendant and victim) appear. A broad distinction is made between those social actors that have an agent role-the doer of the process who is given some degree of agency (that is, the grammatical subject of an active voice verb); and those that have a recipient role-the "done-to" actors who lack agency or are the beneficiaries of the process (i.e. the traditional direct and indirect object of a non-passive verb or the subject of a passive voice verb). I excluded those cases that are peripheral to the actual process (as in The defendant placed the bomb near this boy), as they rarely occur and are part of prepositional phrases, rather than processes. Also excluded was direct reported speech, as this form of speech representation is, objectively speaking, not overtly mediated by the author's interpretation. All the processes identified were then coded based on Halliday's framework (1994) above, and the frequency counts were then normalized to a common basis, per 1,000 clauses, to allow for comparison of different text lengths. In cases where a process is realized periphrastically (as in He wanted to cause damage), I recognized this as consisting of two processes (namely, a mental and a material process). However, in the case of periphrastic auxiliaries, which function more grammatically (as in He is going to plant a bomb), I identified the process of the chaemsaithong MANUSYA: Journal of Humanities 24 (2021) 204-226 main verb (here, plant a bomb as a material process). The analytical procedure is illustrated in (1) and (2) below. Note that, in the examples, the prosecution is shortened to "Pro", and the defense to "Def", after the name of the case.  (1), there are two clauses. In both clauses, the defendant is given full agency carrying out material processes, with a victim as the recipient of those processes in both clauses. In (2), out of 5 clauses, the defendant appears as the agent of two verbal processes, and as a recipient of a mental process.
Going beyond sheer frequencies, the qualitative analysis was finally performed to investigate how specific process choices position the defendants and victims ideologically. In the qualitative analysis, agency is conceptualized as existing along a continuum or a cline (O'Connor 2000; Dreyfus 2017), ranging from a participant who is more active, more powerful, more personally and morally involved in a process at one end, to a participant who is passive and reacts to other more active, powerful agents at the other end. Three levels of agency are distinguished. These distinctions are summarized in Table  2. In the first column, the lawyer represents the defendant's action as caused by another agent, thereby deflecting agency through the passive voice. In the third column, the lawyer attributes, through the active voice, full agency and responsibility to the defendant in a negative and positive way. Alternatively, as the middle column exemplifies, the defendant may be shown to have partial agency in an act by using the active voice, but he or she does not bring about an effect on others. Or, while the defendant's involvement in the crime is acknowledged, it is also explained away in terms of motivation or reason. The notion of agency as existing along a continuum is a useful concept, as it does not view agency in a binary way (i.e. having agency vs having no agency), and thus it will be adopted in this study and will be used to explain the patterns in the qualitative analysis.

Overall Frequencies
The overall frequencies for the defendants, displayed in Appendix 1, indicate that the defendants occur in the agentive role more frequently than in In terms of the process types, the defendants appear most frequently as agents of material processes, followed by relational, verbal, and mental processes. There are also some behavioral processes, while existential processes are not found. In comparison to the agentive role, the recipient role attributed to the defendants is much less frequent in both sides' speech, suggesting they are not affected by other actors. One notable exception exists, however. In Malvo's case, the defendant is cast in the recipient role at a high frequency rate (216/1,000 clauses). An initial observation that can be drawn at this point is that the prosecution's speech centers on the defendant's involvement in the events in question and their ability to affect others, while the defense does not make such issues the focus of their discourse.
References to the victim, presented in Appendix 2, also indicate contrasting linguistic patterns between the two sides. Across the data set, the prosecution's talk features a high presence of the victims, as agents as well as recipients of processes, while the victims are scarce in the defense's talk. This means that the prosecutor foregrounds the involvement of the victims in the events in question as well as victimization, while the defense attempts to background these issues. There is, however, one exception. In comparison to other cases, the defense in Murray's case appears to attribute more agency to the victim, Michael Jackson, than does the prosecution. As a result, the defense can position him as actively involved in and responsible for causing his own death. In terms of process types, when they appear in the discourse, the victims are mostly represented as being affected by, or executing, material processes (but, as will be shown shortly, without an affected participant). This is followed by relational, verbal, mental, and behavioral processes.

4.2
Transitivity Patterns for the Defendants The qualitative findings indicate that the prosecution attributes full agency to the individuals on trial. The predominant pattern appears to be that of the defendants as autonomous agents executing various effectual actions. This unmitigated agency is enhanced through the use of "would" to indicate an iterative pattern of activity in (3a), impactful material processes with direct objects in (3b), and an emphatic verb phrase ("made absolutely no attempts") in (3c). As a result, each defendant is held morally responsible for what happened to the victim. The defense, in contrast, proffers a different account of reality by attributing mitigated agency to the defendants. That is, these individuals are represented as capable of performing certain material processes, but incapable of affecting other participants. This is illustrated in (4a), where the defense describes Malvo's childhood before meeting Muhammad. The defendant is represented as an agent who can walk, get to school, change school, and drop out of school, but note that he is a rather ineffectual agent, at least in comparison to the way he is represented in (3b). There are also times when the defense creates a polarized identity of the defendant by casting him as an effectual agent, but only in a benevolent way, as shown in (4b) Verbal processes also appear to be ideologically significant. More than half of the verbal processes attributed to the defendants show authorial evaluation on the part of the prosecution, portraying the defendants as incriminating themselves or lying to the interrogating officer. In (5a), for example, Darlie emerges as untrustworthy ("gave differing stories" and "claimed"); in (5b), the lawyer not only attributes the admissive statements to the defendant but also uses a paralinguistic verb ("laughed") to suggest the defendant's unremorseful attitude; in (5c), the defendant is represented as concealing the truths ("makes no mention", and "Michael was physically and emotionally fine"). Oftentimes these reported verbal utterances are followed by outright invalidation of the reported content.
(5a) This defendant gave differing stories about what had happened... Neither the defendant, nor any intruder ever ran through that kitchen and utility room to the garage after the boys' attack, as the defendant claimed. That never happened...no intruder ever threw a knife down on that utility room floor...as the defendant claimed...no intruder ever left out of that garage through a window... That never happened either, as this defendant claimed it happened that night. (Darlie Pro) (5b) The first one-he said the first one he killed that morning-was the lawn mower guy... Later in talking about it, this defendant laughed. He laughed. He said, "The lawn mower kept going after the guy went down". (Malvo Pro) (5c) Conrad Murray's own tape-recorded interview, he makes no mention of using the ambu bag; He said things to Kenny Ortega when concerns are expressed, he said things such as "I am the doctor, not you. You direct chaemsaithong MANUSYA: Journal of Humanities 24 (2021) 204-226 the show and leave Michael's health to me". He said "Michael was physically and emotionally fine. Don't let it be your concern. I am the doctor". (Murray Pro) With respect to other process types, similar representational patterns are observed, as shown in Table (3). In Darlie's case (6a), the prosecution depicts the defendant as facing problems related to her children, which allegedly motivated her to murder them, whereas the defense constructs a world where the defendant has a warm family and shows a glimpse of her internal feelings to portray her as a caring mother. In Malvo's case (6b), all the processes that the prosecutor presents point to the defendant's share in the criminal acts, whereas the defense silences the killings and focuses on Malvo's malleability, naiveté, and his bond with Muhammad. Finally, in Murray's case (6c), the prosecution downgrades Dr. Murray's medical identity to a service provider, who would do anything for money and deauthorizes his expert identity as a cardiologist, while the defense authorizes his medical identity as well as expertise.
When the defendant is cast in a recipient role, it is mostly for each side to supply more details that corroborate the theory of the case. For example, Dr. Murray's defense team creates a story in which their client was a recipient of Jackson's verbal processes in order to attribute the cause of Jackson's death to the singer's persistent complaints about his sleep problems and his requests for a more powerful sleep aid. Similarly, in Routier's trial, the defendant becomes a "victim" of the investigating team in (7a). The prosecution, in contrast, portrays Dr. Murray as a mercenary doctor, and so Dr. Murray is cast as the potential recipient of money in exchange for his service (7b).
(7a) Dr. Murray told the detectives about Michael Jackson's problems...in the spring of 2009 when Michael Jackson came to him and asked him for help...Michael Jackson told Dr. Murray that he had inability to sleep and the only way he could sleep was on propofol and that he had always slept on propofol when he was touring. Dr. Murray told the investigators, this took him back. This didn't make any sense to him. Michael Jackson told Dr. Murray how propofol was applied, he told Dr. Murray that it needed to be used with lidocaine, he told Dr. Murray that he had a nickname for propofol...Michael Jackson told Dr. Murray that you had to give it with Lidocaine otherwise it would burn...

Transitivity Patterns for the Victims
In comparison to the defendants, the victims appear much less frequently in the opening addresses of both sides. The presentation of the victim as recipient occurs frequently in the prosecution's speech, and this is, to a great extent, related to the presentation of the defendant. That is, the prosecution's emphasis on the defendant as an effectual agent requires the presence of a direct recipient of the defendant's processes.
It was observed that the defense's strategy is to draw attention away from the victims and that the impacts the victims sustain are backgrounded or even silenced, thereby concealing the responsibility of the defendant, as shown in (9), through nominalized processes (e.g. "horrendous act of destruction"), removal of the agent in the passive voice (e.g. "A Bushmaster was used" and "Did the person that got shot", use of a process without participants (e.g. "the events begin" and "there's a robbery). When the victims are allocated an agentive role, the prosecution positions the victims in the subject position of material processes without a goal, thereby mitigating their agency. This is for the purpose of describing how they lived or what they were doing before the crime, as shown in (10). As a result, the defendant's attacks appear motiveless and ruthless. Note that this is in contrast to the strategy we saw earlier in the presentation of the defendant as the agent, in which case the defendant appears in transitive clauses. Interestingly, in Routier's case, the prosecution presents the victims as the impactful agents to explain the defendant's motive to kill her own sons, although we must bear in mind that these agents do not exercise control over their actions, as they are children. Moreover, the intention here is not to distribute responsibility and blame to the grammatical agents. Rather, because the agents are the defendant's children, the defendant is depicted as violating the parental norm of caring for one's children.
(11) That baby...keeping her busy.... Two children ... were also keeping her busy... was taking more and more time... (Routier Pro) As mentioned in the quantitative section, Murray's case appears to present an exception to the defense's main strategy of victim representation. In this particular case, the defense assigns responsibility to Jackson himself for causing his own death, hence proffering an opposite version of reality by presenting him as the willful agent of various processes, as in (12) and (13).
(12) On two occasions, Michael Jackson told Dr. Murray "I have anxiety and I have trouble sleeping". Michael Jackson had a problem that no amount of determination, no talent would ever overcome. And he knew that he needed help. In the spring of 2009 Michael Jackson went out looking for that help. (Murray Def) (13) We will ask every single expert that the prosecution presents, everyone, this question, "If the drugs that Michael Jackson received on the 25th was exactly what Dr. Murray said, would that have killed Michael Jackson? Everyone of them will tell you no, we believe. And every one of them will tell you that there had to have been more propofol delivered,

Conclusion
This study has examined how agency and responsibility are negotiated through lexico-grammatical resources in the opening speech. This research reveals that two opposing opening statements are coherently accomplished by attributing different levels of agency to the defendant and the victim(s) in each case. In particular, the prosecution tends to assign full agency and autonomy to the defendant, so that he or she appears as a determined agent of processes that ultimately lead to the alleged unlawful act that the defendant is being tried for. In contrast to the defendant, the victim is attributed limited or no agency, so that he or she appears as an agent who does nothing wrong and lives his or her life, and yet is affected by the defendant's acts.
The defense, in contrast, creates an opposite identity of both the defendants and the victims. Like the prosecution, the defense may attribute full agency to the defendant, but differs in that this agency is for the benefit of other participants. Alternatively, the defense may passivize the agency of the defendant, so that he or she appears as victimized by others, rather than affecting others. As a result, the defendant's unlawful acts are backgrounded or dismissed as not true. With regard to the victim, the defense is mainly silent about the experience of the victims that emerged as a result of the defendant's acts. This is achieved by excluding the victim from the clauses (either through processes that require no participants or through nominalization), or by attributing agency to the victim so as to suggest his or her share in the unlawful act.
In a more critical vein, this study explicates how the opening statement is not a mere "road map" that provides an overview of the case and introduces the evidence to be presented later on. Instead, as the third research question suggests, the opposing grammatical representations are indexical of the presenters' stances and function to evoke positive and negative assessments in the jurors. Because the text-wide tendencies in transitivity choices have the potential to dispose the jurors toward a particular attitudinal assessment, the opening statement becomes not only argumentative but also an evaluative representation of people and their actions. That these opening statements are not recognized as violating the legal standards perhaps is attributable to the absence of explicit subjective intervention by the lawyer. This study, thus, raises a critical issue of the implicitly evaluative mechanisms, in the form of grammatical choices, that are indexical of the presenter's ideological positions, through which particular points of view and value orientations can be made to seem to arise naturally from what appears on the surface to be factual informational content. It is hoped that the findings of this study may have practical implications for training jurors and the public at large to be aware and more critical of the power of grammar in constructing and shaping courtroom experiences. A future topic to be pursued is to also consider the roles of reference terms (such as the defendant, last name, first name, title, etc) that lawyers employ together with transitivity patterns in the creation of competing representations of the same social actors.