A Pragmatic Analysis of Questions and Responses in American Defence Attorneys' Cross-examinations

The structure of the interrogation process in crossexaminations is said to be diverse and complex in terms of question-response typology. This is because the counsel has to extract truth from an opposing party’s witness whose views are expected to advocate that party's views regarding the case. Accordingly, the study which is basically quantitative in nature aims to investigate what the examining party intends to obtain out of these questions and which of these questions are the most prevalently used. It also aims to measure the amount of cooperativity in witnesses' responses. Accordingly, three transcripts of cross-examination have been analyzed, using a pragmatically-oriented approach. The approach draws on Stenstorm (1984) and Archer's (2005) classification of questions; Stenstorm (1984) and Archer's (2002) classificatory scheme of responses which is based on the strategies of violating Grice's (1975) maxims to determine the degree of cooperation on the part of respondents. The analysis revealed a diversity in the attorneys' method, making the use of four types of leading questions as well as non-leading ones represented by WH questions. The latter recorded the least percentage in comparison with the overall percentage of leading questions. That is; a preference is shed on the part of cross-examining counsel towards leading over non-leading questions. Moreover, the majority of the responses given have indicated the witnesses' commitment to the purpose and format of the questions posed, showing a high level of cooperativity on the part of those witnesses.


The Introduction
Language is crucial to the judicial system; it is a tool by means of which law is interpreted in different cultures. Research interest in the use of language in courts and other institutional settings has flourished so rapidly over the past decades as documented. This is because the study of language is essential to achieving a better understanding of the judicial process (Levi, 1990).
Linguistics is among the scholarly disciplines which contribute significantly to the study of language in the world of law. It provides valuable insights into the nature and interpretation of the language used in any legal system. For this distinctive form of language associated with law, Kwarcinski (2019) opted for the term legal discourse as a certain type of language which largely depends on how utterances function within a communicative situation which cannot be separated from its social context. Kwarcinski (2019) believed that legal texts consist of utterances whose form and structure "fit most comfortably into the realm of linguistic inquiry" (para. 2). However, he further added at the same time that the framework of one discipline is not enough to analyze legal discourse. Such components as function and context involve extra-linguistic factors; therefore, he called for the necessity of analyzing the two components of legal discourse, i.e., text and context, in an interdisciplinary enterprise. Hence, he suggested that the study of legal discourse should, in addition, be associated with that particular area of linguistics called pragmatics.
In terms of the pragmatically-oriented approach, Kwarcinski (2019) viewed legal discourse as consisting of contextualizing utterances. That is; a text serves a particular communicative purpose which "corresponds to the author's perlocutionary intention to change the legal situationan intention manifested in the text and reflected in the actual illocutionary force of its component utterances" (para. 1).
Moreover, in viewing legal discourse as a text which occurs in a particular context, one can account for its internal diversity, This is because it is possible to recognize within it different genres. Each has a particular secondary communicative purpose apart from the primary communicative purpose common for legal discourse in general. One such genre is 'crossexamination' which is the stage of a trial during which examining counsels, and sometimes, judges engage in a well-planned process designed to interrogate witnesses. Their major aim is to prove their case by eliciting testimony from the examined witness who has just testified for the opposing party in an attempt to reveal inconsistencies in that witness's testimony or to cast doubt on its accuracy during this stage of a trial.
Though questions and answers are the only kinds of turns that the turn-taking system in cross-examinations is confined to, the structure of this interrogation process in cross-examination is diverse. The diversity leads to have complexity in terms of question-response typology. This further reflects the complexity of the situation where the attorney has to extract truth from an opposing party's witness whose views are expected to advocate that party's point of view regarding the case. Hence, it is hypothesized that question acts in cross-examinations are not only limited to leading questions, but also to nonleading ones. The variety in the use of question forms helps to test the reliability of the adverse evidence, and to elicit from the witness any favorable evidence that helps the examiner advance their own case. Answering questions precisely when cross-examined by the opposing attorney is what witnesses should do according to the traditional cross-examination procedures. Therefore, witnesses' responses should be to the point; they should not elaborate on them. Rather, their responses should be within the strict limits set by the questions. However, it is hypothesized that witnesses appear sometimes to be evasive by not being to the point or by giving less or more information than is required, violating one or more of Grice's (1975) maxims.
To know about the nature of this question-response typology, the present study is to examine how the cross-examining counsels in cross-examinations structure their questions in the interrogation process. To meet this objective, the researcher needs to: identify the different types of the question acts used, the most prevalently used ones in the data selected, and the information that the examining party intends to obtain out of such question acts taking into account their illocutionary force. Moreover, it aims to examine the structure of the responses given to measure the conformity of the responses with the questions asked and to find out how cooperative a witness is in his/her response. It is worth noting that the data selected consists of three transcripts of cross-examinations from which (300) sets of questions and responses have been excerpted for the purposes of analysis.
In linguistics, significant contributions to research on the language of law as a linguistic phenomenon have been made. Such contributions traced its evolutions and examined the characteristics of its vocabulary and sentence structure taking into account the text as the basic unit. In addition, almost all previous research was conducted on the interface of text and context to examine the legal implications of the language of law in cross-examinations as a subgenre of legal discourse. The main focus of such research was on the act of questioning through which cross-examining counsels practice their power dominance to force witnesses to provide a response. However, little research has been conducted from a purely pragmatic perspective to examine the act of asking a question and responding to it as the basic interactional unit in this legal domain. This study is an attempt to fill in this gap.

Cross-examination
According to judges, cross-examination is a necessary and sufficient method in terms of which a great deal of trial evidence is confronted.
Its importance as an essential portion of the trial necessary to test the truth of direct examination is stressed by Stritmatter (2009). The latter added, it is not only a science which has developed its recognizable rules, techniques and methods, but also an art which helps assemble the artistic components of cross-examination. However, its circumstances are different from those of direct examinations where the witness is rehearsed by the trial attorney and directed to how to act in accordance with the goals set by the attorney. Atkinson & Drew (1979) adopted rules of conversational analysis to examine the organization of the types of courtroom speech exchange. They believed that the same rules govern the talk in courts. However, courtroom talk differs from ordinary conversations in that certain constraints operate in the former. For example, in turn allocation rules, the speakerchanges are regulated, especially the kind of talk which occupies the greatest part of the time allocated for the examination of witnesses and defendants in most courts; namely, crossexamination. Thus, an examination is characterized by two important features making it markedly different from conversations. In examinations, turn order and the type of each speaker's turn are both fixed due to the fact that the talk in examinations involves only two parties whose turns are designed as either questions or answers. In contrast to conversations, the interrogative syntactic structure of a speaker's utterance does not guarantee its status as a question. Therefore, utterances in conversations are not necessarily treated as questions or answers. Moreover, the two types of speaker turns in examinations are not randomly distributed. Rather, they are preallocated in that only one party operates the techniques for allocating the next turns through the production of questions and self-selection of the following answers until the completion of the examination. In other words, only the examiner who conducts the examination has the right to ask questions. The utterances produced by the examined party should always be answers positioned sequentially after the examiner's questions. This further entails that the party being examined cannot exercise the right to ask the examiner a question. Hence, the examination is characterized by having the order A-B-A-B as opposed to conversations which need not be ordered in this way (Atkinson & Drew, 1979).

2.2.Question-Response Typology
Both questioning and answering are an art and technique; however, there have been many shifts in emphasis over time regarding which of these two techniques fundamentally create synergies in human thinking, action, and interaction. Experts in the field of linguistics have long committed themselves to examining the different forms and functions of different kinds of questions used for varying purposes in various circumstances. Moreover, several theoretical schools in linguistics have been interested in the description of the distinctive features of questions in general and of their many uses in particular (Ilie, 2015). Questioning as a speech act solicits an answer which may or may not be successful. For an answer to be successful, it must satisfy different criteria, such as: appropriateness, cooperativeness, informativeness, and usefulness (Kiefer, 1988, as cited in Athanasiadou, 1994. However, it may not succeed, and the reasons are many including the following: "The recipients may not hear or understand the talk, they may ignore it and continue to be involved elsewhere or initiate other actions, they may hear and understand but withhold their responses" (Pomerantz, 1984, p. 152, as cited in Athanasiadou, 1994). Whether or not they are successful, the answers are different techniques which differ not only because of the different situations of use, but also because of the different types of questions.

Typology of Questions
In his discussion of the types of questions, Ilie (2015) talked about controversies in the definitions and classifications of questions. He maintained that it is impossible to agree upon a definition that combines formal, interactive, and pragmatic criteria. Different types of criteria have been made use of to classify questions occurring in natural languages in several ways. Archer (2005) adopted the best way to classify questions; it was proposed by Quirk et al. (1972) who classified them via their syntactic forms into Yes/No interrogatives, alternative interrogatives, tagged declaratives, and WH-interrogatives according to the type of answers they elicit. However, the classification of questions in terms of the type of answer expected was criticized in favor of the claim that this classification should instead be based on formal criteria. This is because such a classification clarifies the reason that alternative interrogatives and WHinterrogatives both elicit the same answer. Ilie (2015), on the other hand, necessitated the application of both semantic and pragmatic criteria to highlight more subtle aspects of the intended messages which questions convey. Taking into account a set of pragmatic criteria such as the appropriateness of answers to questions, the types of speech acts enacted, and the relation between the illocutionary force of questions and their perlocutionary effect, a distinction can be made between standard and nonstandard questions. Standard or genuine questions are so called because they fulfill the basic and most common function of questions which is expecting an answer. As such, standard questions encompass all answer eliciting and information-eliciting questions. Nonstandard questions are those which fulfil functions other than eliciting information or explicit answers such as expressing a challenge, an invitation, a reproach, a complaint, a warning, a threat, an objection, a protest, an accusation, etc. Such functions are all context-specific depending on interlocutors' roles, their power positions, goals, and relationships between them. Nonstandard questions comprise a wide range of diverse questions, such as: examination questions, riddle questions, rhetorical questions, and echo questions, which occur in different institutional and non-institutional settings.  Fuller (2003) believed that the speech act of questioning in cross-examination involves different syntactic forms which the attorney can use to achieve various ends. Thus, WH-questions allow witnesses to be free in their answers. Unlike Yes-No questions, they involve a limited range of answers the witnesses can possibly give. Declaratives ending with a question tag are even more powerful and more inclined to indicate veiled accusations. For Rubinowitz & Torgan (2002), WH-questions are non-leading questions, which they think, should never be asked on cross. This is because they are open-ended unless they are low-risk questions to which the attorney knows the answer or the answer does not cause him/her any damage. Therefore, they called for Yes/No questions, which in their opinion, are leading questions as these contain the answer within them, or rather suggest the answer. Leading questions are mostly Yes/No questions, but they can also take the form of a statement to the end of which the words "true," "correct" or "right" are added. Or, they can be formed by adding phrases like "would you agree with me," "isn't it true," "is it fair to say," "isn't it a fact" or "there is no question that" to the front of the question. No such words and phrases are added if the examiner produces the statement in a way that makes it sound like an interrogatory. Whatever their forms are, leading questions serve one purpose which is to list the facts that the examiner would like to prove through the witness. Moreover, through leading questions, examiners can impeach the witness with a prior inconsistent statement that s/he testified on direct examinations. Besides, they can commit them to their prior testimony by attacking it immediately. In his quantitative research on the pragmatic properties of question types, which crossexamining attorneys manipulate at trial, Woodbury (1984) found out that leading questions, which restrict the scope of the witness's response to either 'yes' or 'no', are the most frequent types of questions that the attorneys use allowing them to retain the floor. Thy reduce the scope of the witness's response and his/her role to a minimal respondent. In a case study carried out by Tkačuková (2010) on the power of the act of questioning in crossexaminations, she discussed those general principles that are applicable to different genres of institutional discourse. Then, she proceeded to examining the discursive tools that are available to cross-examining counsels to limit witnesses' responses through questioning. Tkačuková (2010) comes to the conclusion that "closed questions are the primary means used by counsels. However, they also used pragmatic tools that require training and experience" (p. 59). Tkačuková (2010) added that these pragmatic tools "range from lexical means (a careful choice of words used in questions …) to prosodic means or even turn-taking management (ironic tone or prolonged pauses after significant replies can be used strategically to convey additional meanings …)" (p. 53). However, in another study, Kizito and Kondowe (2017) used the theory of Conversation Implicature supplemented by Halliday's interpersonal metafunction to examine the power of questioning exercised by the prosecution during court trials. They found out that s/he was able to dominate the courtroom discourse and control the witnesses when they asked them non-leading questions represented by Wh-questions. Such questions demanding a minimal response which "gave no room for new information apart from the state's anticipated response" (p. 140). Another tool that was available to the prosecution to enact power apart from their dominant use of questions is their choice to structure their sentences in the imperative mood to issue commands and request.
Whether s/he asks leading or nonleading questions, Stritmatter (2009) pointed out that the cross-examiner dominates all aspects of questions, their number, and the speed with which these questions are put to the witness.

Typology of Responses
Having identified what a question is, the next step is to identify an answer which requires, in the first place, an identification of the positioning of questions. For Archer (2005) however, this is not adequate enough to identify an answer as questions may sometimes be followed by another question, which is why he raised the question: What is an answer? In this regard, Groenendijk & Stokhof, (1984) remarked that answers are attempts on the part of the respondent to fill in gaps signaled by the questioner in her/his information. They added that pragmatics must be involved in an essential manner, for their belief that the semantic interpretation associated with the questions is not sufficient to determine the answerhood thesis.
According to Rubinowitz & Torgan (2002), the attorney should always be in good control of the witness. Moreover, s/he should review and use impeachment materials to elicit the answers they require and to reveal inconsistencies. Rubinowitz & Torgan (2002) listed types of answers the witness may give, saying that the answer "may be totally unresponsive, partially responsive with gratuitous information added or totally responsive. As a general rule, the trial attorney should settle for nothing short of a totally responsive answer" (p. 3). The attorney can deal with an unresponsive witness in two ways, either directly by showing the jury how uncooperative the witness is or indirectly by repeating the same question until the witness gives a responsive answer.

The Adopted Model s
The analysis focuses on the question/response typology of the adjacency pair as a basic interactional unit of crossexaminations in the aforementioned three cases. For this step, an eclectic model has been adopted. It is based on Stenstorm (1984) and Archer's (2005) classification of questions in terms of their forms and interactional intent and the responses expected. It is also based on Stenstorm (1984) and Archer's (2002) classificatory scheme of responses in terms of question forms. The latter scheme further depends on the strategies of violating Grice's (1975) maxims, and it helps to determine the degree of cooperation on the part of those respondents.

Stenstorm (1984) and Archer's (2005) Classificatory Scheme of Questions
In her classification of question acts, Stenstorm (1984) made an initial distinction between primary question acts whose major function is to elicit and re-elicit a response and secondary ones which function as complements.
Taking into account what the questioner wants the respondent to do, she recognized four groups of primary question acts. For example, the type of question act required if the questioner wants the respondent to supplement details, which the [(Re-) elicit], lacks must be Q: clarify and Q: repeat as illustrated in Table ( As she classified primary Q acts in terms of the type of response given, Stenstorm (1984) admitted that the lexico-grammatical form of the question determines its function. That is; a WH-question, for example, functions mainly to 'identify' to which the respondent is expected to identify the referent of the WH-word. As shown in Figure ( type of Q act to her classification of questions in terms of the form/function relationship. It is 'Q: acknowledge', i.e., 'request for acknowledgement', which occurs at the borderline between the question and the statement. In his approach which he developed following Stenstorm (1984), Wierzbicka (1987) and Carletta et al.'s (1997) categorization of question functioning, Archer (2005) assigned utterances to one or even more of seven macro categories. Such categories "are meant to exemplify the main instances of verbal behavior in the (historical) courtroom" (p. 128). Among these macro categories are questions which subsume a set of values including: -Ask (about) as when S seeks to get A to confirm something about Y <Q: confirm>, -Inquire (into) as when S solicits a verbal response from A causing the former to know something about Y and is divided as follows: ask for a polarity decision <Q: polar> and ask for a missing variable <Q: identify>, question/ascertain as when S solicits a verbal response from A in a systematic way <Q: classify>, -Interrogate as when S uses force to elicit a verbal response from A <Q: control>, -Query as when S seeks clarification of what A says regarding Y <Q: clarify>, and -Entreat as when S seeks to get A's verbal permission for y <Q: permit>. Answers are placed into the most appropriate macro category regarding the force field which tends to be 'inform' in most cases. However, additional sub-fields to express the various functions of answers together with their corresponding questions are also included in this category, as shown in the following section. It is worth noting that Archer (2005) designed his macro categories together with the values they subsume in order to explicate the involvement of the speaker (S) and addressee (A) and their sensitivity to such social variables as status, role, and distance.

Stenstorm (1984) and Archer's (2002) Classificatory Scheme of Responses
Taking into account the type of response expected, Stenstorm (1984) believed that the syntactic form of the question in addition to its lexical features and intonation are, sometimes, unreliable criteria as they do not always guarantee its function. Thus, typical responses given to Q: identity, Q: polar, and Q: confirm represented by WH-questions, Yes/No questions, and declarative questions, respectively would be 'identify', 'affirm/deny', or, 'confirm/disconfirm'. However, there are those instances when WHquestions function indirectly as Q: action (request to perform an action), Yes/No questions as Q: offer, and declarative questions as Q: permit (request to grant permission). In case one, it is more likely to expect R: comply, R: accept, etc. as other contextually appropriate responses.
Among the taxonomies devised to examine the nature of responses given to questions is that of Stenstorm (1984). The latter's categorization of responses relies upon 'continuation options' in that the demand on the respondent to give a response is conditioned by the question eliciting force which varies in terms of its function. Therefore, a request for information demands an obligatory response unlike a request for acknowledgement which demands an optional response. Yet, it is not always possible to predict whether or not a response will follow or what that response will be. Rather, what the respondent will do in the utterance, which follows regarding the effect of a certain question, remains a matter of expectation. Table (2) illustrates Stenstorm's (1984) representation of the degree of response expectation and eliciting force in relation to Qtype: In Archer's opinion (2005), part of the function of the answer as a basic interactional phenomenon in the courtroom is to inform or provide information. Following Harris (1984), Philips (1984), and Stenstorm's (1984) insights into the nature and functioning of answerhood, Archer (2005) listed a number of possible values which the answer subsumes as shown below: -Answer as when S responds verbally in a way which wholly or partially fills the gap with the required knowledge that A is seeking; -Respond as when S responds to something said previously. It differs from 'answer' in that what S says in response to something is not built in expectedly; -(In)validate as when S provides a polarity decision explicitly; -Identify as when S provides required information explicitly; -Imply as when S does not say 'yes', 'no', or provide a value for the missing variable explicitly but gives an answer in such a way that it can be inferred; -Supply as when S provides information which is not required because s/he does not possess the exact information; -Elaborate as when S provides more information than is explicitly requested. It includes amplify (when S emphasizes the information s/he gives), qualify (when S provides additional information), and expand (when S immediately builds on the information given); -Confirm as when S says that Y is true; -Oppose as when S says that Y is false; -Disclaim as when S shows her/his compliance but unable provide the requested information despite her/his; -Evade as when S shows non-compliance so s/he does not say 'yes' or 'no' or does not provide a value for the missing variable, and does not give an answer in a way that can be inferred; and -Refuse to answer as when S in not totally complied refusing to give a verbal answer even though it is required. It is worth mentioning that Archer (2005) describes both evade and refuse as "conscious avoidance manoeuvres" (p. 341). Walker (1987) believed that a responsive answer for both linguist and lawyer alike must satisfy Grice's (1975) four conversational maxims. Therefore, it should be as informative as is required and only informative (Maxim of Quantity), truthful (Maxim of Quality), relevant to the question (Maxim of Relation), and spoken in a clear, brief and orderly manner (Maxim of Manner). In brief, an answer must be one which provides a direct and precise response to the question which usually has, in a legal setting like the courtroom, a command function forcing the respondent to reveal her/his knowledge in an appropriate manner.

Archer's (2002) Scheme of Response Description Using Grice's (1975) Maxims
In Archer's (2005) opinion Grice's (1975) four maxims "can help us to identify those occasions when respondents adopt a less-thandirect approach as a means of thwarting the examiner's line of argument" (p. 57). That is when the respondents do not give the type of response typically expected by the examiner's question. Hence, they either tend to opt out or violate the maxims using one of four strategies including: fabrication, concealment, evasion, and vagueness. Such maxims include the following: Fabrication, as indicated by McCornack (1992), involved "distorted versions of the sensitive information" (p. 9), as when respondents intend to provide false or inadequate evidence in an attempt to mislead violating the Maxim of Quality.
Concealment is considered a strategy of deception which results from the violation of the Maxim of Quantity. The violation is made when respondents deliberately withhold critical information; i.e., they provide insufficient information or, sometimes, more information than is required in a particular situation with the intention to mislead (Carson, 2010).
Evasion, as defined by Dillon (1997), is "a routine strategy for responding to a question without answering it" (p. 126) by intentionally avoiding to convey "correct/accurate information to manipulate the situation to the speaker's advantage" (Zhang, 2011, p. 577). This happens especially when the information received does not fit his/her expectations (Fraser, 2010). In doing so, the respondent violates the Maxim of Relation.
Vagueness is the act of providing noninformative responses which lack completeness and clarity with the intention to confuse others violating the Maxim of Manner (Ruzaite, 2007). Vagueness can either be 'active' as when respondents intentionally use vague language, and 'passive' when they utilize it having no other alternative because of their lack of knowledge (Zhang, 2011).
Accordingly, Archer (2002) devised the following classificatory scheme which, (as cited in Archer, 2005), is based upon the Gricean (1975) maxims to make a distinction between those instances when a surface level cooperation is exhibited on the part of the respondents. That is; when there is no intention to generate an implicture. It further distinguishes instances when, for some reason or another, they violate one or more of Gricean (1975) maxims. In addition, Archer's coding system includes an 'opt out' category for those instances when a respondent opts out of observing one of the maxims subsumed under the Cooperative Principle. On the other hand, the 'ambiguity' category is for those instances when it is not feasible to identify which maxim has been violated, as illustrated in Table 3: An outline of the eclectic model that has been adopted for the corpus analysis in the present study is presented in Figure 2.

Sample of the Study
In this paper, the data is gathered from a collection of texts available online. The texts selected are transcripts which are accessed from the web-site http://defensewiki.ibj.org/index.php/sample_Cros s-Examination_Transcripts.
They represent samples of defense attorneys' crossexaminations of state experts which are categorized as follows: (1) cross-examinations of forensic experts, (2) cross-examinations of informants, and (3) cross-examinations of sexual assault cases. Following are the three cases which have been selected from the aforementioned categories, respectively: Case 1: Cross-Examination of a Medical Examiner Case 2: Cross-Examination of a Government Cooperator Case 3: Cross-Examination of a Sexual Assault Victim It is worth noting that the corpus used for the purpose of analysis in this study represents a total of (100) cross-examination questions and responses excerpted from each of these three cases, totaling (300) sets of questions and responses.

Statistical Means
To achieve the first objective of the study, a simple computational technique has been used to:  examine which of the different types of question acts identified in the selected corpus are the most prevalently used;  calculate the total number of question types;  compute the percentage of each question type with reference to the total number of these question types.  compute the interactional intents of each question type.
To meet the second objective, which reads: measuring the amount of cooperation in witnesses' responses, the researcher is to do the following steps:  Calculating the total number of responses given to each question type;  Calculating the percentage of cooperative and non-cooperative responses in accordance with the total number of these responses; and  Counting the cooperative and non-cooperative response types.

Typology of Questions
Based on the analysis of 300 questions excerpted from the three selected cases, two types of questions were identified, leading questions and non-leading questions. The former basically tell the respondent what the answer should be. They include declarative questions, Yes/No questions, declarative + tag questions, and alternative questions. The latter type of question, on the other hand, non-leading questions, is represented by WH-questions. As statistically illustrated in Table ( (104) and (102) have a prevailing occurrence with almost equal proportions amounting to 34.66% and 34%, respectively. These results are in line with what is usually recommended that the principal mode of the examination of an adverse party; i.e., the cross-examination, should be framed in such a way that the witness's response should only be confined to 'yes' or 'no'. Such a framing helps verify the statements they produced earlier in the direct examination. In what follows are some excerpts to illustrate the use of these two forms of questions as the most prevailing ones in the data selected: -Q. "All right. That was --when you say a couple, just two?" B. "Yes." -"Did you notice that on the right side of the upper nose, there was an one eighth inch superficial red abrasion?" -Q. "You made a contract with Future Services which was separate from Maalouf?" A. "yes, sir." (Cross Examination of a Government Coordinator, n.d., pp. 154, 156) -A. "Okay. He was home around 6:00?" B. "Yes, sir." -Q. "Prior to the incident you described, had there been any abusive behaviour from your father towards you?" B. "No." (Cross-Examination of a Sexual Assault Victim, n.d., pp.27-8) WH-questions achieved somehow a good percentage allowing it to be in the third place with a sum of (45) questions making up 15% out of the total. They allow the adverse witness to repeat the defence's theory of the case in their own words. This may further result in the examiner's loss of control; therefore, there have been calls to avoid using this form of nonleading questions. Nevertheless, the WHquestions detected were found to be of the type which limits the scope of the response given. That is; respondents were limited to give specific responses identifying only the referents of 'Who?', 'What?', 'When?', and 'Where?' as made clear in the excerpts below: -Q. "What are endorphins?" A. "They are naturally occurring substances that are --react to the nerve endings. And they act like a drug to block the sensations." (Cross-Examination of a Medical Examiner, n.d., p. 86) -Q. "What date did you sign the agreement?" A. "25 October 2007." (Cross Examination of a Government Coordinator, n.d., p. 155) -Q. "How would that work?" A. "My dad worked till like 6:00 at night, and my mom was a teacher at that time, she teaches pre-school, so she was home earlier than he was." (Cross-Examination of a Sexual Assault Victim,n.d.,p.27) In the fourth rank come declarative questions, which were mostly tagged with 'right' and 'correct' onto the end. This implies also a certain desired response, as shown in the following excerpts. This form of leading questions almost exhibits the same frequency of distribution as that of WH-questions with 44 utterances, making up 14.66%. -Q. "So there would be limitations on the amount of activity, correct?" A. "Right." (Cross-Examination of a Medical Examiner, n.d., p. 26) -A. "You left to go back to the U.S. in March, right?" B. "I did, sir." (Cross Examination of a Government Coordinator, n.d., p. 168) -Q. "And, you weren't afraid he was going to kill you on that trip, were you?" A. "No." (Cross-Examination of a Sexual Assault Victim, n.d., p.65) Alternative questions constitute the least proportion in the data with only five instances representing 1.6% out of the total. That is why this type of leading questions and the responses given to them are not taken into consideration in the analysis below. As for the classification of leading and non-leading questions in terms of their interactional intent; namely: declarative questions, Yes/No questions, WH questions, and declarative + tag questions, the results which are statistically detailed in Table (5), revealed that 98 declarative questions were asked with the examiners' intent to limit the witnesses' role to confirming their declarative statements. Therefore, 'confirmation' achieved the highest percentage with 92.30% in comparison with the interactional intents of other declarative questions asked. These results are again consistent with what is normally recommended that lawyers should frame all their questions "so that the witness only says 'yes'" since "the lawyer wants the witness to validate all of the factual statements" (Saylor and Small, 2017, p. 2).

Table 4 A Taxonomy of Question Types in the Three Cases
Yes/No questions, on the other hand, were asked mainly with the intent to elicit a 'yes' or 'no' response where the witnesses are nudged to make one of two choices, either affirming or denying the question. Hence 'polar' was the interactional intent of 89 yes/no questions, making up 87.25%. Such a percentage represents the highest compared with interactional intents of other questions of this type. These results reflect the normal tendency on cross-examinations. That is; for the witness to answer either 'yes' or 'no' as opposed to what many lawyers recommend, the range of answers given should even be narrower than this, limiting the respondents to an affirmative response only. Though WH questions should never be asked during cross-examinations in accordance with the rules, almost all WH questions detected in the three cases were asked with the interactional intent 'identify'. This results accounts for 97.7% with only one WH question (i.e., 2.2%) which was intended to elicit content information from the witness. These results reflect caution on the part of the examiner to formulate such a type of questions in a way which does not allow the witness to give a long response so as not to lose control over him/her. Therefore, they intended their WH questions to elicit short responses which are limited to identifying the time or place or what people or things are.
As for declarative + tag questions, these were 100% designed with the examiners' interactional intent to communicate a bias towards acknowledging what was posed in the questions committing the respondents to their propositional content. In what follows are some excerpts illustrating the interactional intent of each form of the questions used in the data selected: -Q. "But you're not surprised it's not in the dictionary." A. "No." (Cross-Examination of a Medical Examiner, n.d., p. 76) -Q. "Was this anything to do with your work with Future Services?" A. "No, sir." (Cross Examination of a Government Coordinator, n.d., p. 155) -Q. "And, where are the bedrooms?" A. "Upstairs." (Cross-Examination of a Sexual Assault Victim, n.d., p. 24) -Q. "You used the word that there were six stab wounds that were potentially fatal, correct?" A. "Yes." (Cross-Examination of a Medical Examiner, n.d., p. 68) It is worth noting that the interactional intent of the question determines the type of response given. That is why the responses in the extracts mentioned above are considered cooperative as they reflect the respondents' cooperation on the surface level. For more details about this type of responses see Section 3.3.2.1.

Typology and Description of Responses
A close inspection of the corpus selected reveals that responses fall into two categories. The first category is the cooperative responses which represent respondents' cooperation on the surface level. That is; they are non-misleading and direct in that they provide the information that is directly requested by the examiner's question. The second category is the noncooperative responses which indicate respondents' non-observance of certain maxims or their unwillingness to give a cooperative response, i.e., they opt out. Statistically, constitute the vast majority of the responses given to each of the four major types of questions detected whereas non-cooperative questions constitute the lowest percentage given to each type of these questions.

Cooperative Responses
As explicated in Table (7) below, the vast majority of the responses the witnesses gave to declarative questions were either to re-confirm or disconfirm what was conveyed in the question. Treating these declarative questions as merely seeking re-confirmation or disconfirmation, these unexpanded responses were found to consist mostly of the interjection 'yes', making up 79.80% in contrast with disconfirming responses, which constituted only 2.88%. These results are evidence of the examiner's desire to limit the witness's responses to confirming the declaratory questions.
Similar results were obtained as regards the responses given to Yes/No questions, the second major type of the leading questions detected. Most of these responses came in the affirmative form totaling 50% in addition to 3.92% of them which were in the confirmative form in contrast with 'deny' responses which make up 17.64% only. These results are compatible with the goal of the examining party, which is to elicit an affirm/deny response from the witness "to verify the truthful facts sought to be established by the cross-examiner" (Dodd, 2012, p. 6).
In accordance with the interactional intent of WH questions posed by the examining party to elicit short responses focusing only on specific contextual information, the witnesses seem to comply with the constraints this form of nonleading questions imposes on them. They imply giving brief information that conforms to the constraining nature of these questions making up 97.77%.
Finally, 88.63% of the responses are given to declarative + tag questions. They represent surface level cooperation on the part of the respondents, conforming as a result to the interactional intent of this type of questions which is to acknowledge the information provided in the question.

Non-cooperative Responses
Sometimes, however, witnesses were observed to give responses to declarative questions. Such responses involved more than a mere confirmation or disconfirmation by giving additional information, clarifying, or elaborating after the interjection. Or, they might just provide information without even giving a confirm or disconfirm response. These expanded noncooperative responses which altogether constitute 17.30% violate the Maxim of Quantity. This is because the respondents' contribution was more informative than is required. One reason for this mismatch in the form of response given to the question posed is perhaps the witness's desire to give more details on the subject. Or, it might be due to the idea that neither the 'yes' or 'no' response reflects their own opinion on that subject.
As for non-cooperative responses given to Yes/No questions which amount to 28.43% as a whole, it was also found that witnesses were more likely to provide content information in some of these responses, making up 21.56%. Again, they violated the Maxim of Quantity by giving more information than is required despite the fact that this form of leading questions does not allow witnesses to explain, clarify, or elaborate. This is attributed to the belief that the examiner may lose control over the witness. The remaining responses which encompass such expressions as 'I don't know', 'I don't remember', and 'it is a possibility' constitute 5.88%. Following the categorization of hedging presented by Prince et al. (1982), such expressions are referred to as 'shields', which "change the relationship between propositional content and the speaker by implicating a level of uncertainty with respect to speaker's commitment" (as cited in Fraser, 2010, p. 7). Failing to give a precise response represented by either 'yes' or 'no' to this type of questions, respondents violated the Maxim of Quantity by giving less information than what is required. Examinees' imprecise responses are perhaps indicative of their fears that they may be accused of being wrong. There was only one evasive response (0.98%) when the respondent in Case 2 refused to affirm or deny what was posed in the question as explicated below: -Q. "Do you remember in your head if you were lying or telling the truth then?" A. "Sir, I've spent many hours, at least 20 hours, going over stuff. And it's beenthere was a lot of stuff going on during that time period. To recall one sentence or one statement is a lot to ask for in something like that." (Cross-Examination of a Government Cooperator,n.d.,p. 190) Again, his evasive response is evidence of his fears that choosing an 'affirm' or 'deny' response might have negative legal implications. Only one response given to WH questions was found to diverge from the constraints imposed by this type of questions. It constitutes only 2.22% when the examinee in Case 2 failed to contribute sufficient information as illustrated below: -Q. "When did he hire you?" A. "I don't remember the exact date." (Cross-Examination of a Government Cooperator, n.d., p. 155) In using the hedging expression "I don't remember", the examinee violated the Maxim of Quantity.
As for the responses to declarative + tag questions, it was found that 9.09% of them were associated with more excessive information than is required, violating again the Maxim of Quantity. Only one response (2.27%) represents a violation of the Maxim of Manner when the plaintiff in Case 3 gave a vague response as shown below: -Q: "But, you didn't think he was going to seriously hurt you for that, right?" A. "I wasn't sure." (Cross-Examination of a Sexual Assault Victim, n.d., p. 65) According to Prince et al. (1982), the expression 'I wasn't sure' is "speech act hedging" which "serves as an index of the commitment of the speaker to the truth of the propositional content conveyed" (as cited in Fraser, 2010, p. 6). To the examiner, however, a response like 'I wasn't sure' is vague since vague expressions involve those which lack precision as expected by the examiner.

Conclusions
Unlike the prevailing belief that crossexaminations should consist only of leading questions, the examination of the structure of three cases in question has revealed a diversity in the method followed by the attorneys. Both leading and non-leading questions were used to prove the consistency of what the witnesses said on direct examinations. A mix of leading and non-leading questions were identified in the corpus selected. Non-leading questions, represented by WH-questions, recorded the least percentage in comparison with the overall percentage of the four types of leading questions: declarative questions, Yes/No questions, declarative + tag questions, and alternative questions. This shows a preference on the part of examiners for leading over non-leading questions. This is because the former is more likely to elicit the responses wanted due to their forms which force the examinees to provide a specific response. Taking into account what is most likely the examiner's intent in asking each type of questions, it seems that the need to get a confirmative response is strongly sensed in the majority of declarative questions asked. The need to elicit an affirmative or negative response is also detected in the examiner's strong preference to ask Yes/No questions. Therefore, the majority of them are asked with the interactional intent 'polar'. To acknowledge the truth of what is posed in the examiner's question is also the desired interactional intent behind all the declarative + tag questions. In short, the interactional intent of these three major types of leading questions reflects the examiners' desire not to allow for details or elaborations regarding the questions raised. This also seems to be their desire when asking WH questions as it is mirrored in the illocutionary force of this type of non-leading questions. The purpose of such questions was to urge the respondents to just identify specific times, places, people, or qualities.
On the other hand, examinees exhibit cooperative behavior as is evidenced from the responses given. Thus, the majority of the responses to leading questions indicate the respondents' commitment to the purpose of this type of questions which is simply to get a 'yes' or 'no' response. Respondents seem also to abide by the format of non-leading questions which requires them to be specific in their responses. Therefore, almost all their responses are constrained to only identifying the referent of the WH word. Only a small amount of noncooperativity is detected as is clearly shown by the minority of the responses. The latter seem to deviate from the norm either by giving more or less information than is required, using hedging expressions, or by refusing to answer the questions posed to them.