WITNESS RESPONSE MANIPULATION THROUGH STRATEGIC NON-LEADING QUESTIONS (OR THE ART OF GETTING THE DESIRED ANSWER BY ASKING THE RIGHT QUESTION)

WITNESS RESPONSE MANIPULATION THROUGH STRATEGIC NON-LEADING QUESTIONS (OR THE ART OF GETTING THE DESIRED ANSWER BY ASKING THE RIGHT QUESTION) I. INTRODUCTION Sydney A. Beckman* It s not what you say, but how you say it. Origin Unknown Words, whether spoken or printed, are the world s primary form of communication. 2 This communication may come in person, from a letter, over the telephone, through a computer or smart phone, via email, through video chat or even a text message. 3 Although body language, facial expressions, and context also impact communication, they generally are merely supplements to the primary form of communication. Words chosen by the speaker (or author) can have a significant impact on the listener. This is particularly true when it comes to questions. * Professor of Law, Lincoln Memorial University Duncan School of Law. Professor Beckman would like to thank his Research Assistant Deanna Breeding for all of her hard work; Professor Katherine Marsh for tackling some difficult challenges quickly and efficiently; Ann Long for finding that which could not be found; and his wife, Allyson, for her constant and unwavering support. 2. See definition of communication. Webster s Third New International Dictionary 460 (1961). 3. Communication can be divided into two broad categories: verbal and nonverbal. While verbal communication is an exchange of words, nonverbal communication consists of behavior such as head nods, facial expressions, gestures, posture and eye movements, as well as vocal variety including pitch, tone, pace, and volume. See Michael Argyle, Florisee Alkema & Robin Gilmour, The Communication of Friendly and Hostile Attitudes by Verbal and Non-Verbal Signals, 1 EUR. J. SOC. PSYCHOL. no. 3, 1971, at 385, 387. 1

2 SOUTHWESTERN LAW REVIEW [Vol. 43 Questions do more than ask: they solicit and convey information, and focus and suggest answers. By influencing answers, questions alter what is understood by others. 4 The quote that began this introduction, It s not what you say, but how you say it, is often repeated; it is used in speeches and referenced in jokes. In the context of adversarial proceedings, the use of words has rarely been analyzed with regard to the formation of questions and, more importantly, the impact of those words on the answers provided. Is the accuracy of answers impacted by the question posed? Does an answer to a question depend, in part, on the structure, intonation, dialect or speed used to form the question? This article discusses the question of whether or not the words, specifically in the formation of questions, impact the answer thereby permitting the questioner to manipulate the responses. 5 Further, the article will briefly touch on other facets of questioning such as: temporal impact on answers; variables such as age, gender and verbal cues; inattentional blindness; and how these facets may impact answers. Additionally, the use of punctuation as it relates to the written word and the impact of punctuation on the message being communicated will be briefly discussed. Finally, this paper will make suggestions as to how to combat these manipulative techniques in the courtroom. II. WHY WORDS MATTER What are words for - when no one listens anymore? Missing Persons Dr. Kathy Kellerman posits that questions shape answers in four ways. First, the wording of a question puts words in answerers mouths. 6 Second, that [q]uestions frame acceptable answers, suggesting some and excluding others. 7 Third, [q]uestions carry assumptions that influence answers. 8 Finally she suggests that [a] question s form influences 4. KATHY KELLERMANN, PERSUASIVE QUESTION-ASKING: HOW QUESTION WORDING INFLUENCES ANSWERS 5 (2007), available at http://www.kkcomcon.com/doc/kpqa.pdf. 5. Although the scope of this paper is limited to an examination of the impact on responses, other research has evaluated the impact of questions on juries and their perceptions. See, e.g., Saul M. Kassin, The American Jury: Handicapped in the Pursuit of Justice, 51 OHIO ST. L.J. 687, 692-97 (1990); Saul M. Kassin, Lorri N. Williams & Courtney L. Saunders, Dirty Tricks of Cross- Examination: The Influence of Conjectural Evidence on the Jury, LAW & HUM. BEHAV., Aug. 1990, at 373, 373; John C. Reinard & Darin J. Arsenault, The Impact of Forms of Strategic and Non-Strategic Voir Dire Questions on Jury Verdict, COMM. MONOGRAPHS, June 2000, at 158, 158. 6. KELLERMANN, supra note 4, at 1. 7. Id. at 2. 8. Id.

2013] WITNESS RESPONSE MANIPULATION 3 answers by inviting agreement or disagreement, openness or evasion, and threat or comfort. 9 At times, questions may inadvertently mislead fact finders. Oftentimes a listener may be confused by, or simply may not be paying close enough attention to, a question and/or the subsequent answer. In one study, mock jurors listened to an excerpt of testimony and indicated whether certain statements were true or false. After hearing the statement, I ran up to the burglar alarm, for example, most subjects recalled that the witness had said, I rang the burglar alarm. 10 Kassin surmises that people may process information between the lines and assume they heard what was only implied. 11 Words and their impact on responses can, under certain conditions, be readily understood. Consider the following polls that were conducted during the Vietnam War: President Johnson ordered the bombing of Hanoi and Haiphong. Two members of Congress conducted independent surveys of the issue in their respective districts. Member A posed the question as follows: Do you approve of the recent decision to extend bombing raids in North Vietnam aimed at the strategic supply depots around Hanoi and Haiphong? Sixty-five percent [(65%)] favored the decision. Member B posed the question as follows: Do you believe the U.S. should bomb Hanoi and Haiphong? 12 In response to Member B s wording, only fourteen percent (14%) favored the decision. 13 It may be easily hypothesized why these two questions, although seeking the same information, received significantly different responses. In the former question, the responder could even if incorrect surmise the basis for the bombing thereby giving the responder more information to make a determination. In the latter question, less information is given to the responder. This may lead to the response being based on a number of factors but, of significance, gives the responder no basis from which to surmise the reason for the bombing. Under Kellermann s different categories, the Vietnam questions likely fall under her fourth option. That is, the first iteration of the question 9. Id. at 3. 10. Kassin, The American Jury, supra note 5, at 693 (emphasis omitted). 11. Id. 12. Charles S. Reichardt, Wording of Questions (2004) (unpublished manuscript) (on file with author). 13. Id.

4 SOUTHWESTERN LAW REVIEW [Vol. 43 possibly invited comfort or a reason for agreement. The second question lacks such comfort and reason. Other scenarios may be less overt as to the impact on responses and many factors may influence answers. It has been suggested that the most prominent external influence on recollection is the power of suggestion.... 14 In 1974, the question about words and their impact on responses, was asked by two researchers at the University of Washington. 15 In this study the researchers sought to discover whether the use of different verbs posed in questions about the speed of a vehicle would elicit different answers and specifically, different answers that, 16 were statistically significant. 17 Asking questions in different ways may translate into effective ways to impact trial results. 18 Word choice will impact answers. 19 In a study 20 of the case of the Commonwealth of Massachusetts v. Kenneth Edelin, 21 terms were examined which were associated with the 14. Justin S. Teff, Human Memory is Far More Fallible and Malleable Than Most Recognize, N.Y. ST. B.J., June 2004, at 38, 40. 15. See Elizabeth F. Loftus & John C. Palmer, Reconstruction of Automobile Destruction: An Example of the Interaction Between Language and Memory, 13 J. VERBAL LEARNING AND VERBAL BEHAV. 585, 585 (1974). 16. The abstract of the article summarizes the purpose of the study more fully: Two experiments are reported in which subjects viewed films of automobile accidents and then answered questions about events occurring in the films. The question, About how fast were the cars going when they smashed into each other? elicited higher estimates of speed than questions which used the verbs collided, bumped, contucted [sic], or hit in place of smashed. On a retest one week later, those subjects who received the verb smashed were more likely to say yes to the question, Did you see any broken glass?, even though broken glass was not present in the film. These results are consistent with the view that questions asked subsequent to an event can cause a reconstruction in one s memory of that event. Id. 17. The idea of an infinite population distributed in a frequency distribution in respect of one or more characters is fundamental to all statistical work. From a limited experience, for example, of individuals of a species, or of the weather of a locality, we may obtain some idea of the infinite hypothetical population from which our sample is drawn, and so of the probable nature of future samples to which our conclusions are to be applied. If a second sample belies this expectation we infer that it is, in the language of statistics, drawn from a different population; that the treatment to which the second sample of organisms had been exposed did in fact make a material difference, or that the climate (or the methods of measuring it) had materially altered. Critical tests of this kind may be called tests of significance, and when such tests are available we may discover whether a second sample is or is not significantly different from the first. RONALD A. FISHER, STATISTICAL METHODS FOR RESEARCH WORKERS 43 (Oliver & Boyd, 1925). 18. The study specifically addressed subsequent questions over a time-differential in an effort to address whether memory would fill in gaps. See Loftus & Palmer, supra note 15, at 585. This too would have implications at trial but exceeds the scope of this article. 19. See KELLERMANN, supra note 4, at 10. 20. Brenda Danet, Baby or Fetus? Language and the Construction of Reality in a Manslaughter Trial, 32 SEMIOTICA, no. 3-4, 1980 at 187. In this article, the scientific method was

2013] WITNESS RESPONSE MANIPULATION 5 victim, the subject of a criminal trial. In that 1975 case, the defendant was tried on a manslaughter charge. The issue, for purposes of that discussion and this article, surrounded terminology used at trial to reference a fetus. On the one hand, the prosecution wanted to personalize the victim while the defense sought to accomplish the opposite. As noted by the author, the charge of manslaughter, or any type of killing, quite naturally takes for granted that the victim was once alive. 22 One might think that the use of select language might make little difference in the outcome. However, so concerned with the use of language was the defense counsel, that he filed a motion for an order to prevent use of the words suffocate, smother, murder, baby boy and human being. 23 The prosecution offered to use the term male child instead of baby boy, both of which were objected to by defense counsel. 24 Do these terms matter? That is, does the use of a particular term over another impact how a jury deliberates? This author often posits that, in a court of law, the truth is what the jury believes regardless of whether their conclusion is based on what factually occurred. The way words describe, characterize, chronicle, depict, detail, explain and inform are impacted by the words chosen. The words chosen are impacted by the rules and procedures of the judicial system. As noted by Danet, the words baby and fetus are referentially more or less equivalent ways of talking about [the victim] but socially, they are miles apart. 25 The wording of questions can impact the witness in such a way as to impact his or her answer. Danet notes that questioning in the adversary system, with its rules and procedures, results in a reality that is constructed and negotiated in the courtroom.... 26 And, she notes, the outcome of questioning is as much a function of the verbal strategies and choices of the participants as it is of the supposed facts of the case. 27 Kassin points out that questions may mislead a jury (or fact finder) when suggestive questions... produce support for... conjecture by shaping the witness s testimony. 28 not employed but rather transcripts of proceedings were merely evaluated and conclusions drawn from the analysis. 21. 359 N.E.2d 4, 9-10 (Mass. 1976), abrogated by Commonwealth v. Cass, 467 N.E.2d 1324, 1329 (Mass. 1984). 22. Danet, supra note 20, at 187. 23. Id. at 188. 24. Id. at 189. 25. Id. at 191. 26. Id. 27. Id. at 192. 28. Kassin, The American Jury, supra note 5, at 697.

6 SOUTHWESTERN LAW REVIEW [Vol. 43 This conclusion bears out in a number of real-world examples. One such example comes from the General Social Survey (GSS). 29 From data collected over an eighteen-year period, the GSS asked the following questions: Are we spending too much, too little, or about the right amount on welfare? Response Options Percent of Responses Too Little 20 % About Right 33 % Too Much 47 % In the same data sets with responses from the same individuals, the following question was also asked: Are we spending too much, too little, or about the right amount on assistance to the poor? Response Options Percent of Responses Too Little 64 % About Right 25 % Too Much 11 % When reviewing this data, Dr. Charles S. Reichardt surmised, in effect, that the wording of the question asked essentially the same thing and yet, the words used resulted in significantly different responses. Reichardt rhetorically ponders whether politicians tend to use the word assistance when they want to increase funding to the poor but [use] the word welfare when they want to cut back on funding to the poor. 30 In People v. Simpson, the defendant, O.J. Simpson, was being tried for the murder of his ex-wife, Nicole Brown Simpson. 31 In that case, much like the Edelin case, the opposing sides used significantly different words to 29. NAT L OP. RESEARCH CTR., GENERAL SOCIAL SURVEYS, 1972-2012: CUMULATIVE CODEBOOK (2013), available at http://publicdata.norc.org/gss/documents/book/ GSS_Codebook.pdf. 30. Charles S. Reichardt, Wording of Questions (2004) (unpublished manuscript) (on file with author). 31. People v. Simpson, No. BA097211, 1995 WL 704381 (Super. Ct. Los Angeles County Oct. 3, 1995).

2013] WITNESS RESPONSE MANIPULATION 7 describe identical events. 32 For example, the defense called domestic incidents that which the prosecution called wife beating and abuse. 33 Clearly, the former label is intended to sterilize the incidents while the latter is designed to evoke negative emotion towards the defendant. Strategically crafting questions, which in turn manipulates answers, could significantly impact trial results. Seemingly innocent questions, such as one inquiring about a person s height, may not be so innocent after all. III. MISDIRECTION Think of a card, any card... A magician, somewhere in the world, thirty seconds ago How tall was the man running through the alley? This innocent question may, in fact, be more than meets the eye. One would argue that a question such as the one posed is neutral, that is, nothing about the question suggests the answer. Questions which do, in fact, suggest the answer are governed by Rules of Evidence. Federal Rule of Evidence (FRE) 611 34 governs questions which suggest answers in terms of leading questions. The Rule provides, in relevant part, that [l]eading questions should not be used on direct examination except as necessary to develop the witness s testimony. Ordinarily, the court should allow leading questions: (1) on cross-examination; and (2) when a party calls a hostile witness, an adverse party, or a witness identified with an adverse party. 35 But what is a leading question and, more to the point, is a question such as How tall was the man running through the alley? a leading question? The Federal Rules of Evidence do not define a leading question. Black s Law Dictionary defines a leading question as [a] question that suggests the answer to the person being interrogated.... 36 Few would argue that a question such as this suggests, in any way, the answer. If we reframe the question in this way: Would you say the man running through the alley was about six feet tall? then we have, clearly, suggested an answer thereby creating a leading question; one that is potentially governed by FRE 611. 32. See Reid Hastie & Nancy Pennington, The O.J. Simpson Stories: Behavioral Scientists Reflections On The People Of The State Of California V. Orenthal James Simpson, 67 U. COLO. L. REV. 957, 964-65 (1996). 33. KELLERMANN, supra note 4, at 5. 34. All states except New York have enacted a rule that is similar, and in some cases identical, to Federal Rule of Evidence 611(c). See infra Appendix C. 35. FED. R. EVID. 611(c). 36. BLACK S LAW DICTIONARY 969 (9th ed. 2009).

8 SOUTHWESTERN LAW REVIEW [Vol. 43 Now examine the question reframed again. How short was the man running through the alley? This question is virtually identical to the original question merely substituting the adjective short for the adjective tall. A choice of adjectives, in the above examples, does not in either case suggest the answer. That is to say, using the adjective tall does not, overtly, provide guidance to the responding party as what height the interrogator is seeking. Nor, presumably, does the adjective short provide such guidance. Therefore by the strictest definition the questions would not, in the truest sense, be objectionable based on a FRE 611 leading objection. In 1974 a study was published that examined the effect of using various words within questions to determine whether word choice would impact the answer. 37 In this study, a test was conducted to investigate... the phrasing of the question used to elicit speed judgment. 38 The subjects who participated in the study were shown films which depicted a traffic accident. 39 Subsequently the subjects had written questions posed, 40 one of which inquired about the speed of the vehicles in the film. Subjects were divided into groups and the question, examined for the study, varied the word choice as to the adjective used to inquire about the speed of the vehicles. The verbs tested were as follows: smashed, collided, bumped, hit and contacted. 41 The control word, or neutral word, appeared to be hit. Seven different films were shown to different groups in an effort to replicate results with different input. 42 The results may seem surprising. Words which implicated higher speeds resulted in higher estimates. Table 1 43 reflects the results of the study. 37. Loftus & Palmer, supra note 15, at 585-86. 38. Id at 585. 39. Id. at 586. 40. Id. Presumably the use of written, as opposed to oral, questions eliminates variables such as voice inflection, tone, speed, et cetera which arguably could influence an answer. The use of written questions eliminates these such variables permitting the researcher to focus on, and appropriately analyze, the issue being examined. 41. Id. 42. Id. 43. Id. at 586 tbl. 1.

2013] WITNESS RESPONSE MANIPULATION 9 Table 1 Verb Used in Study Mean Speed Estimate Smashed 40.5 Collided 39.3 Bumped 38.1 Hit 34.0 Contacted 31.8 Loftus and Palmer conclude that [t]he results of the experiment indicate that the form of a question (in this case, changes in a single word) can markedly and systematically affect a witness s answer to that question. 44 The researchers noted that the actual speed of the vehicles played little part in the subjects estimation of speed but the phrasing of the question controlled considerable variance. 45 A subsequent experiment was performed which evaluated, in part, how time and question phrasing impacted recall. Part of that investigation offered a slight variation of the above experiment. In this subsequent test, subjects were asked about speed with either the use of the word hit or smashed. 46 Again, a statistically significant difference was found between the responses based on the wording of the question. 47 A different approach was taken in a study which examined the use of what the researcher called marked and unmarked adjectives and adverbs. 48 An unmarked word was one which had a nominal or neutral use that refers to the whole dimension. These were compared with the marked words which had a lower bounds of zero. For example, the names of dimensions such as high versus low were used in questions such as How high is the building? (which would have an answer that is potentially unlimited) versus How low is the building? (which would have an answer that is limited to zero). 49 In this study, subjects were asked to make numerical guesses to questions which contained either marked or unmarked words. The hypothesis being tested was whether or not the use of these marked modifiers would impact responses to a statistically 44. Id. at 586. 45. Id. 46. Id. at 587. 47. Id. Loftus and Palmer offer an opinion as to why the different words result in different responses. Those opinions, while interesting, exceed the scope of this paper. 48. Richard J. Harris, Answering Questions Containing Marked and Unmarked Adjectives and Adverbs, 97 J. EXPERIMENTAL PSYCHOL. 399, 399 (1973). 49. Id. Although not addressed in the study, arguably even use of the question How high is the building? is not as neutral as What is the height of the building? Nevertheless, Harris presumed that use of the word low presents an, although unmentioned, upward boundary.

10 SOUTHWESTERN LAW REVIEW [Vol. 43 significant factor in the direction of the marking while use of unmarked modifiers would not. 50 Using the example of the height of the building, in the context of the hypothesis, the question How high was the building? should yield answers uninfluenced by the question while the question How low was the building? would presuppose that the building was low. 51 In this study, subjects were told that the experiment was a study in the accuracy of guessing measurements and that they should make as intelligent a numerical guess as possible to each question. 52 Think back to the question posed at the beginning of this section: How tall was the man running through the alley? Might an answer be manipulated by phrasing the question How short was the man running through the alley? Appendix A reflects a complete list of the test questions used by Harris, the marked and unmarked modifiers, and the average estimates by the subjects. 53 Table 2 illustrates a select few of the questions: Table 2 Questions and Units of Measure Modifier 54 Mean Response How time did the man have between planes? (min., hr.) much little 73.2 min 37.5 min How money was in the man s wallet? ($, 4) much little $ 4.90 $ 3.96 How was the set of weights? (lb.) heavy 146.3 lb. light 56.3 lb How was the quarterback? (lb.) heavy 185 lb. light 180 lb. How was the office building? (stories) high 26.2 stories low 13.1 stories How was the plane flying (ft. off ground) high low 8,907 ft. 4,481 ft. 50. Id. 51. Id. 52. Id. at 401. The methodology of the experiment was that the examiner...read a list of 32 questions and [the subject] wrote a numerical answer for each. The [Examiner] read each question once, told the [subjects] the unit(s) of measure to be used in their answers, and then repeated the question. Each group of [subjects] received the same 32 question frames in a different random order. Items were counterbalanced such that each [subject] heard every question frame and adjective or adverb only once, with half of the [subjects] hearing the unmarked member in a given question frame and half hearing the marked member. Id. 53. Harris, supra note 48, at 400. 54. The top word is the unmarked word while the bottom word is the marked word. Id.

2013] WITNESS RESPONSE MANIPULATION 11 The results support the hypothesis that answers to unmarked-modifier questions cover a wider range than answers to marked-modifier questions. 55 The table reflects, in some case, enormous variances in responses. For example, a difference of ninety pounds in response to the questions of How heavy were the weights? versus How light were the weights? The results clearly indicate that the wording of a question may affect the answer. 56 Loftus notes that in the context of both past personal experiences and recently witnessed events that the wording of questions may impact answers. 57 Even seemingly minor changes in wording may influence an answer. An interrogator need not be as overt as using language such as tall or short, high or low. In fact, the use of definite or indefinite articles may impact an answer. 58 Use of the word a does not necessarily presume the existence of an object. For example, the question Did you see a broken headlight? does not presume the existence of a broken headlight. 59 However, use of the definite article does: Did you see the broken headlight? In the former, the inquiry is whether one existed or not. This assumes the responder would see it if, in fact, it existed. In the latter, the inquiry is wholly different. In that case, the assumption is that the headlight existed and the inquiry is whether or not the responder saw the headlight. The significance in the inquiry is whether or not the use of the definite article as opposed to the indefinite article would alter a response. The results of a study in which this hypothesis was tested supports the proposition that it does. 60 In that study, Loftus and Zanni found that very small changes in the wording of a question will influence responses. 61 In two experiments, subjects viewed a film of an automobile accident and were subsequently 55. Id. at 401. Harris notes that: [p]rior to figuring the statistics, the numerically largest score in each of the 64 distributions was eliminated, since there were several distributions with one very extreme, though not logically impossible, score, whose inclusion in the sample made the variance meaningless. In addition, three other single scores in three different distributions were eliminated for unintelligibility or [the subject s] obvious misunderstanding of the question. Id. 56. Elizabeth F. Loftus, Leading Questions and the Eyewitness Report, 7 COGNITIVE PSYCHOL. 560, 561 (1975). 57. Id. 58. Id. 59. Id. 60. Elizabeth F. Loftus & Guido Zanni, Eyewitness Testimony: The Influence of the Wording of a Question, 5 BULL. PSYCHONOMIC SOC Y 86, 86 (1975). 61. Id.

12 SOUTHWESTERN LAW REVIEW [Vol. 43 asked questions about what they saw. 62 For some of the questions, the use of the definite article the was used while in others, the use of the indefinite article a was used. 63 Previous literature had been divided in their conclusions. 64 Loftus and Zanni note that implicit in the question Did you see a broken headlight? are two questions. First, was there a broken headlight and, second, if there was a broken headlight, did you see it. 65 This logic dictates that if the responding party decides that, in fact, there was a broken headlight then he or she should be fairly certain of his or her response to the second implicit question. 66 The researchers conclude that the problem that arises for a subject is that filmed accidents occur in the space of seconds (much like the reality of eye-witness testimony) making it nearly impossible to be certain of [the first question thereby resulting in a response of] don t know much of the time. 67 Contrasting this scenario with use of the definite article creates a wholly different inferential chain. When the question is phrased with the use of the definite article: Did you see the broken headlight? the first question referenced above, was there a broken headlight, is completely eliminated. Loftus and Zanni hypothesized that fewer don t know responses would result when the definite article was utilized. 68 In the experiment, 69 the subjects were informed that they were participating in an experiment on memory and would be shown a film and 62. Id. 63. Id. 64. Loftus and Zanni report that Muscio (1915) concluded that the more reliable form of question was one that did not use the definite article, whereas Burtt (1931) reported that a and the are about equally suggestive. Id. 65. Id. at 87. 66. Id. 67. Id. 68. Id. 69. [M]ethod: One hundred graduate students participated in this experiment, in groups of various sizes. All subjects were told that they were participating in an experiment on memory and that they would be shown a short film followed by a questionnaire. The content of the film was not mentioned. The film itself depicted a multiple-car accident. Specifically, a car makes a right hand turn to enter the main stream of traffic this turn causes the cars in the oncoming traffic to stop suddenly, causing a five car bumper to bumper collision. The total time of the film is less than 1 min, and the accident itself occurs within a 4-sec period. At the end of the film, the subjects received a questionnaire asking them to first give an account of the accident you have just seen. When they had completed their accounts, a series of specific questions was asked. Six critical questions were imbedded in a list totaling 22 questions. Half the subjects received critical questions in the form, Did you see a....? and the other half of the subjects received them in the form, Did you see the....? Three of the critical questions pertained to items present in the film and three to items not present. Subjects were urged to report only what they saw, and did so

2013] WITNESS RESPONSE MANIPULATION 13 subsequently asked to complete a questionnaire about the film. 70 Three questions inquired about objects present in the film and three questions inquired about objects not present in the film. 71 The results were significant. Regardless of whether an item was actually present in the film, subjects presented with a question using the indefinite article a were over twice as likely to respond I don t know. 72 On the other hand, subjects to whom questions were posed that used the definite article the tended to commit themselves to a yes or no response. 73 Table 3 reflects the percentage of Yes, No, and I don t know responses to whether items were, or were not, present in the film: Table 3 74 [Object] Present [Object] Not Present Response the a the a Experiment I Yes 17 20 15 7 No 60 29 72 55 I don t know 23 51 13 38 Experiment II Yes 18 15 20 6 No 62 28 69 56 I don t know 20 57 11 38 The Object Present column demonstrates responses to the queries when the object about which the question was directed was, in fact, present in the film. The right hand Object Not Present column reflects responses when the object was not present. As noted by the researchers, a major finding was that questions containing an indefinite article led to many more I don t know responses. 75 When an indefinite article [ a ] was contained in a question about an item that was not present in the film, yes responses occurred 6% of the time. When the definite article [ the ] was used, yes responses occurred 20% of the time. I don t know responses occurred, overall, more by checking yes, no, or I don t know. Each subject received a different permutation of the questions. Id. 70. Id. 71. Id. 72. Id. 73. Id. 74. Id. at 88 tbl. 1. 75. Id. at 88.

14 SOUTHWESTERN LAW REVIEW [Vol. 43 often when the indefinite article was used (47.5% vs. 15.5% for the definite article). 76 The conclusions support the hypothesis that the words used to form the question do, in fact, influence the answer or response. The source of this influence may, according to Loftus and Zanni, be the result of two possible scenarios. First, they suggest that the definite article influences the responding party by producing a bias which favors a yes or no response. 77 The second possible explanation they propose is that the definite article leads a subject [the responding party] to infer that the object was in fact present, causing for some a reconstruction in their original memory for the event. 78 In an unpublished study, Loftus interviewed 40 people about their headaches. 79 Loftus found a statistical significance in the difference between answers based on the questions posed. For example, individuals reported more headaches when asked about the number of frequent headaches versus the number of occasional headaches. 80 These results have been replicated in varying contexts. In the study conducted by Southeastern Louisiana University, the adjectives used in the phrasing of a question were examined. 81 The participants were asked to estimate speed of the cars involved in a collision on a 5-point scale. 82 Half of the participants were interrogated using the adverb fast as in How fast was the car traveling? versus the adverb slow as in How slow was the car traveling? 83 The hypothesis being that the use of the adverb slow would impact the value of the response by being suggestive in nature; i.e., the car must have been traveling at a slower speed than that which might 76. Id. 77. Id. 78. Id. 79. Loftus, Leading Questions, supra note 56, at 561. It should be noted that the details of the study were not reported and this author makes no representation as to whether a group of 40 subjects is enough for reliability. Nevertheless, the study does provide useful data for the dialogue presented in this paper. 80. Id. 81. Thomas J. Lipscomb, Hunter A. McAllister & Norman J. Bregman, Bias in Eyewitness Accounts: The Effects of Question Format, Delay Interval, and Stimulus Presentation, 119 J. PSYCHOL. 207, 208 (1985). In this study, 180 students (90 males and 90 females) were recruited from introductory psychology classes. The subjects viewed three representations of an automobile accident. The sequence showed a collision between two vehicles at an intersection. Subsequently, a close-up view of both vehicles was shown. With regard to the portion of the experiment which tested changes in phrasing, words were substituted with marked or unmarked descriptors: fast versus slow. Id. at 209. 82. Id. at 209. 83. Id.

2013] WITNESS RESPONSE MANIPULATION 15 have been otherwise perceived. 84 the study. Table 4, below, reflects the findings of Table 4 85 Questions and Unit of Measurement Modifier Mean Response fast 36.20 How was Car 1 going? (mph) slow 27.50 fast 3.62 86 How was Car 1 going? (scale) slow 2.56 fast 39.76 How was Car 2 going (mph) slow 35.25 fast 3.52 87 How was Car 2 going (scale) slow 3.31 much 6.27 88 How damage was done to Car 1? little 5.40 How were the skid marks made by Car long 5.04 89 1? short 3.88 How noise occurred as a result of the much 7.44 90 accident? little 6.52 How bruises do you think the driver of many 5.22 91 Car 1 suffered? How bruises do you think the driver of Car 2 suffered? few many few 3.85 5.27 92 4.47 The results support the hypothesis. Estimations of the speed of both cars were significantly greater when the unmarked adverb fast was employed as compared to the marked adverb slow.... Similarly, estimates of the extent of damage, skidding, noise, and harm to occupants were all 84. Id. at 208-09. 85. Id. at 210 tbl. 1. 86. Values could range from 1 to 5 on a likert scale (from very fast to very slow). Id. at 10 n.a. 87. Values could range from 1 to 5 on a likert scale (from very fast to very slow). Id. at 209-10 n.b. 88. Values could range from 1 to 11 on a likert scale. Id. at 209-10 n.c. 89. Id. 90. Id. 91. Id. 92. Id.

16 SOUTHWESTERN LAW REVIEW [Vol. 43 significantly greater when the relevant questions were phrased with unmarked... adverbs. 93 In a recent study of nearly 1,500 people 94 which examined the wording of a question, 95 Jay Olson found that the exact wording of the question seemed to influence the decisions made by the responders. 96 Simply changing a single word in the directive, name a card versus visualize a card, impacted the responses. 97 Just as descriptions paint pictures in the minds of the listeners, so do questions which seek a response. As Kassin points out, [c]arefully chosen words can obscure and even alter people s impressions, as when tax increases are called revenue enhancements, and the strategic defense initiative is referred to as star wars. 98 Regardless of the reason, the results of these studies are consistent; the wording of the question can, in fact, impact the response. As mentioned earlier, the reality may be one that is constructed and negotiated in the courtroom.... 99 IV. MISINFORMATION No matter how big the lie, repeat it often enough and the masses will regard it as the truth. John F. Kennedy There is a term frequently used when discussing data analysis and computers: garbage in, garbage out. The term is intended to reflect that computers merely process the information provided and if inaccurate or 93. Id. at 209-10. 94. Jay A. Olson, Pick a Card, Not Any Card, GENII MAG., Aug. 2012, at 76. 95. In this study, Olson and two other researchers conducted an experiment and applied well-known techniques from vision science to measure how well people see, remember, like, and choose each of the 52 cards in a standard [American] deck....to measure choice... over a thousand people [were asked] to either name or visualize a card, then [their selections were recorded.] Jay Olson, Revealing the Psychology of Playing Card Magic, SCI. AM. (July 31, 2012), http://www.scientificamerican.com/article.cfm?id=revealing-psychology-playing-cardmagic. 96. Id. 97. [W]hen asked to name a card, over half of the people chose one of four cards: the Ace of Spades (25%), or the Queen (14%), Ace (6%), or King (6%) of Hearts.... But when asked to visualize a card, people seemed to choose the Ace of Hearts more often. In [the] sample, they chose it almost twice as often when asked to visualize (11%) rather than name (6%) a card. Id. 98. Kassin, The American Jury, supra note 5, at 692. 99. Danet, supra note 20, at 191.

2013] WITNESS RESPONSE MANIPULATION 17 other misinformation is input (garbage in) then the result will be based on such inaccurate or misinformation (garbage out). 100 This concept, albeit in different terms, has been examined by researchers in varied contexts. The results of these studies certainly have application in the adversarial system. There are different ways in which to impact or manipulate an answer. One way is to use words that provide information in a framework which results in completely inaccurate answers. A study consisting of four different experiments examined the suggestion (or input) of false information and subsequently examined the subjects recollections of an event (hereinafter the Loftus Study ). 101 In the first experiment in the Loftus Study, subjects were asked a question in one of two ways: How fast was Car A going when it ran the stop sign? How fast was Car A going when it turned right? The focus of the question was the speed of the car. The significant difference in the question is the presupposition that there was a stop sign. The question presumes the existence of a stop sign. Half of the subjects received the first variant of the question and the other half the second variant. In a subsequent question, all subjects were asked [D]id you see a stop sign for Car A? Fifty-three percent (53%) of the subjects receiving the first variant of the question (the stop sign question) answered yes to the inquiry of whether or not there was a stop sign while only thirty-five (35%) of the subjects receiving the second variant of the question answered 100. Formally, the term is defined as follows: [T]his saying points out the fact that a [computer] can do only what it is programmed to do and is only as good as the data it receives and the instructions it is given. If there is a logical error in [software], or if incorrect data are entered, the result will probably be either a wrong answer or a system [crash]. E.D. HIRSCH, JR., JOSEPH F. KETT & JAMES TREFIL, THE NEW DICTIONARY OF CULTURAL LITERACY: WHAT EVERY AMERICAN NEEDS TO KNOW 595 (3d ed. 2002). 101. The abstract provides that: [A] total of 490 subjects, in four experiments, saw films of complex, fast-moving events, such as automobile accidents or classroom disruptions. The purpose of the experiments was to investigate how the wording of the questions asked immediately after an event may influence responses to questions asked considerably later. It is shown that when the initial question contains either true presuppositions (e.g., it postulates the existence of an object that did exist in the scene) or false presuppositions (e.g., postulates the existence of an object that did not exist), the likelihood is increased that subjects will later report having seen the presupposed object. The results suggest that questions asked immediately after an event can introduce new not necessarily correct information, which is then added to the memorial representation of the event, thereby causing its reconstruction or alteration. Loftus, Leading Questions, supra note 56, at 560.

18 SOUTHWESTERN LAW REVIEW [Vol. 43 yes. The difference of eighteen percent (18%) is statistically significant. 102 Loftus concludes that [t]he wording of a presupposition into a question about an event, asked immediately after the event has taken place, can influence the answer to a subsequent question concerning the presupposition itself, asked a very short time later, in the direction of conforming with the supplied information. 103 In this experiment the existence of the stop sign was, in fact, true. Loftus offers two postulates to explain the result. The first is that when a subject answers the initial stop sign question, he somehow reviews, or strengthens, or in some sense makes more available certain memory representations corresponding to the stop sign. Later, when asked, Did you see a stop sign....?, he responds on the basis of the strengthened memorial representation. 104 Loftus offers a second theory she calls construction hypothesis. 105 When answering the initial question with the presupposition of a stop sign, Loftus hypothesizes that the subject: [V]isualize[s] or reconstruct[s] in his mind that portion of the incident needed to answer the question, and so, if he accepts the presupposition, he introduces a stop sign into his visualization whether or not it was in memory. When interrogated later about the existence of the stop sign, he responds on the basis of his earlier supplementation of the actual incident. In other words, the subject may see the stop sign that he has himself constructed. This would not tend to happen when the initial question refers only to the right turn. 106 Loftus goes on to note that this hypothesis has significance: If a piece of true information supplied to the subject after the accident augments his memory, then, in a similar way, it should be possible to introduce into memory something that was not in fact in the scene, by supplying a piece of false information. 107 Loftus did not reconstruct this particular experiment without a stop sign present. However, a year before, Loftus and Palmer 102. Statistically significant means that the result is unlikely to have occurred by pure chance. See FISHER, supra note 17, at 43. In this case, according to Loftus, there is a less than.05 % chance that the result was by pure chance. Loftus, Leading Questions, supra note 56, at 560. 103. Loftus, Leading Questions, supra note 56, at 564. Although it may not impact Loftus conclusion, it is important to note that Experiment 1 was conducted as a 10-item paper questionnaire as opposed to oral interrogation. Id. at 563. 104. Id. at 564. 105. Id. 106. Id. This hypothesis, however, may have a flaw. Because in this experiment the stop sign did exist, the research does not seem to include a control group or, put another way, a method to determine whether or not subjects answer yes merely had better memories than those who answered no. 107. Loftus, Leading Questions, supra note 56, at 564.

2013] WITNESS RESPONSE MANIPULATION 19 conducted an experiment in which false information was interjected into the questions. In that particular experiment, 108 subjects were shown a film of a car accident and subsequently asked questions by means of a written questionnaire. 109 Subjects were initially asked questions regarding their observation of the speed of a vehicle involved in a collision. 110 The initial questions used verbs of either smashed or hit to characterize the collision. A week later, they were asked questions about the film (without viewing the film again). Subjects were asked, Did you see any broken glass? which were responded to by checking either yes or no. 111 In this case, there was no broken glass (garbage in). The hypothesis was that since broken glass is commensurate with accidents occurring at high speed... the subjects who had been asked the smashed question might more often say yes to the test question. 112 Both tests yielded statistically significant results. With regard to the question of speed of the vehicles, subjects interrogated using the verb smashed estimated vehicular speeds greater than those subjects interrogated using the verb hit. 113 Additionally, as to the second test, subjects who were interrogated with the verb smashed were more likely to answer yes to the question Did you see any broken glass? (garbage out). 114 Other variables may lead to misinformation such as the lack of a shared meaning. Seemingly simple concepts may have distinctly different interpretations. For example, people have interpreted smoking cigarettes as anything from (a) taking even a single puff to (b) cigarettes they have finished, and from (c) cigarettes they have borrowed to (d) only those they have bought. Fully 10% of answers change from yes to no, or no to yes, when given a standard definition of what counts as smoking a cigarette. 115 Kellerman also points out how individual words may carry different meanings to different people. You could mean just me or also my family; 116 weekday might, or might not, include Saturday. 117 Evaluative 108. Loftus & Palmer, supra note 15, at 587. 109. The film lasted less than one minute and the accident lasted four seconds. Loftus & Palmer, supra note 15, at 587. 110. Id. 111. Id. 112. Id. 113. Id. at 587 tbl. 2, 588. 114. Id. 115. KELLERMANN, supra note 4, at 4. 116. Id. (citing W.A. BELSON, THE DESIGN AND UNDERSTANDING OF SURVEY QUESTIONS (Gower, 1981)). 117. Id.

20 SOUTHWESTERN LAW REVIEW [Vol. 43 words, such as not quite and slightly have also shown to be problematic in that they mean different things to different people. 118 Noting that certain words, such as lots, almost all, virtually all, nearly all, a majority of, not very many, almost none, hardly any, a couple, [and] a few generally are interpreted similarly among individuals, other words such as most, numerous, large proportion of, significant number of, considerable number of, [and] several have shown to have highly variable meanings. 119 V. THE IMPACT OF TIME DELAY ON RESPONSES When I was younger, I could remember anything, whether or not it happened. Mark Twain One factor which has been hypothesized as impacting responses is time. 120 [H]uman memory is far more fallible, and indeed malleable, than most recognize. 121 Studies have included as a research variable how time frames impact questions and answers. 122 Questions were asked and answers recorded. Then, sometime later (hours or days), follow-up questions were asked to determine the impact the initial questions would have, if any, on the responses. 123 Some studies failed to demonstrate the ability to mislead or manipulate responses based on questions after a temporal delay. Although not the intent or focus of their study, 124 Yuille and Cutshall found that their attempts to mislead eyewitnesses through the use of biased questions were unsuccessful. 125 They noted, however, that their attempts to mislead came 118. KELLERMANN, supra note 4, at 4 (citing Peter Bartram & David Yielding, The Development of an Empirical Method of Selecting Phrases Used in Verbal Rating Scales: A Report on a Recent Experiment, 15 J. MKT. RES. SOC. 115 (1973)). 119. KELLERMANN, supra note 4, at 4. 120. See Teff, supra note 14, at 41. 121. Id. at 38. 122. See, e.g., Lipscomb et al., supra note 81; Loftus & Palmer, supra note 15, at 586. 123. Although the Loftus and Palmer study used a time delay when testing the question of whether or not a subject saw broken glass, they did not test time a variable and made no conclusion as to the impact of time on their results. Loftus & Palmer, supra note 15, at 587. 124. In the Yuille and Cutshall study, observers witnessed a shooting which resulted in one death and one serious injury. All witnesses were interviewed by the police and a smaller portion agreed to participate in the study four to five months after the incident. The purpose was to evaluate amount and accuracy of recall after the intervening period. John C. Yuille & Judith L. Cutshall, A Case Study of Eyewitness Memory of a Crime, 71 J. APPLIED PSYCHOL. 291, 291 (1986). 125. Id. at 299.

2013] WITNESS RESPONSE MANIPULATION 21 some four to five months after the event and, as such, may have reduced the effectiveness of the manipulation. 126 Time was also a tested variable in a study conducted at Southeastern Louisiana University. 127 Specifically, with regard to the temporal variable, the researchers examined the effect of a twenty-minute delay versus no delay. 128 This study concludes that the delay interval made no significant difference while suggesting that a study involving a longer delay interval might achieve different results. 129 Further research may shed light on the impact these factors would play in a trial setting. Think of the witness who is interviewed, or deposed, prior to a trial that may occur weeks, months or even years later. Depositions are frequently taken relatively early-on in the adversarial process and may be followed by trial testimony sometime later. Grand jury testimony which is elicited with few rules followed by a more structured trial interrogation may be impacted by the temporal effect. More research is necessary to reach a definitive conclusion as to whether or not time would, indeed, impact a witness s responses. VI. HOW DIALECTS MAY IMPACT ANSWERS You like to-may-toes and I like to-mah-toes let s call the whole thing off. George Gershwin Although the premise of this paper is the exploration of how questions are worded, it would be remiss not to mention that fact that other factors such as dialect may also impact responses to questions. Although researchers have not examined dialect per se, in 1978 a study was published which undertook an empirical study to examine presentational style (hereinafter the Conley study ). 130 The Conley study noted that it is possible to distinguish definite styles of speaking and to trace these styles to the social background and the immediate social surroundings of the 126. Id. 127. Lipscomb et al., supra note 81, at 209. 128. Id. 129. Id. at 211. 130. Conley, O Barr, and Lind undertook an empirical study to determine the influence of presentational style on juries functioning as decision makers and analyzes the significance of these findings. John M. Conley, William M. O Barr & E. Allan Lind, The Power of Language: Presentational Style in the Courtroom, 1978 DUKE L.J. 1375, 1377 (1978).