...more common, more likely be inaccurate, more likely to be believed by jurors, and more likely to produce irreversible errors than eyewitness testimony. Nonetheless, the dangers to the innocent posed by conversational testimony have been largely unrecognized. This Article highlights the case for further psychological and legal attention to conversational witnesses by comparing how the psychological processes and legal responses differ between eyewitness and conversational testimony. The Article concludes with implications for reform that may minimize the ongoing and unrecognized miscarriages of justice which result from erroneous conversational testimony.

I. INTRODUCTION

Scholars and practitioners alike share a widespread belief that the single greatest cause of wrongful conviction is erroneous eyewitness testimony. (1) This conventional wisdom is almost certainly wrong. The reason it is wrong is not that eyewitnesses are more reliable than is commonly believed. On the contrary, DNA exonerations have powerfully demonstrated that eyewitness evidence is far less reliable than the system assumes. Of those who have been exonerated by DNA evidence, almost 80% were falsely identified by eyewitnesses. (2) FBI analysis of thousands of DNA samples in eyewitness cases supports the claim that as many as 25% of disputed eyewitness identifications may be erroneous. (3) Rather, the common belief that eyewitness testimony is the major cause of erroneous convictions is wrong for two other reasons: (1) although unreliable, disputed eyewitness identification is rarely involved in criminal prosecutions, and therefore eyewitness error plays a determinative role in only a small fraction of criminal prosecutions; and (2) prosecutions much more commonly rely on evidence that is even less reliable than eyewitness accounts, (4) namely, "conversational" testimony--testimony about what the defendant or his agent said or what was said to him on an earlier occasion. This sort of evidence is present in almost all criminal trials as speech acts (e.g., false or fraudulent statements, offers to buy or sell contraband, bribes, threats), as proof of guilty knowledge (e.g., information imparted verbally to the defendant, statements by the defendant acknowledging awareness of illegality), and as incriminating admissions of the defendant or an agent (e.g., oral statements acknowledging that certain acts occurred or were authorized).

The dangerous inaccuracy of eyewitnesses and the inordinate credence given to them by jurors have been well studied in both legal and psychological literature. In the last five years, there have been more than 400 articles in the psychological literature and 500 articles in the legal literature regarding eyewitness credibility and accuracy. (5) This tremendous amount of attention is due not only to the common belief that eyewitness misidentification is the leading cause of wrongful convictions, (6) but also to the concomitant awareness that an eyewitness mistake often immunizes a guilty perpetrator.

While concern about eyewitness error preceded DNA testing, (7) it has been greatly fueled by DNA exonerations. However, the availability of DNA evidence as a "gold standard" to measure conviction accuracy is mostly limited to violent crimes by unknown perpetrators and, within that small set of cases, to those in which the perpetrator left a biological specimen. While virtually all DNA exonerations involve rape and murder convictions, most crimes are property crimes (e.g., thefts, frauds, forgeries) or "victimless" crimes (e.g., illicit drug transactions, nonviolent sex crimes) in which the perpetrator is either known to the victim or leaves a paper or electronic trail. Even violent crimes are more often committed by acquaintances of the victim where eyewitness identification is not an issue. (8) In fact, a survey of state prosecutors revealed that eyewitness testimony is a central factor in only 3% of felony trials, (9) a figure seemingly consistent with the fact that 16.8% of cases handled by state prosecutors are violent crimes. (10) Virtually all cases in which eyewitness identification is an important issue are those involving violent crimes against strangers. Since most violent crimes are committed against persons who know each other, that leaves less than 9% of felonies in which identification could conceivably be an issue. It seems likely that in no more than one-third of those cases is identification in serious dispute. (11) Wrongful convictions based on eyewitness testimony generate headline attention far out of proportion to their actual incidence.

Testimony about conversations plays a pivotal role in a far more common and broader range of cases, as will be elaborated upon in Part II. It should be noted here, however, that problematic conversational testimony is by no means limited to criminal trials; it extends to securities class actions, (12) interpretation of contracts, (13) trusts and estates, (14) and even guardianship and medical care decision-making. (15) Further, the "rule against hearsay" does not greatly limit prosecutions where out-of-court statements incriminating the defendant are used to convict him. Anything a defendant or his agent said to anyone, at any time, under virtually any circumstance, can, as far as the hearsay rule is concerned, be received into evidence against him in a criminal trial. (16) In addition, a wide array of hearsay exceptions often permits the prosecution to offer hearsay evidence of out-of-court conversations in which neither the defendant nor his agent was even present. (17) For example, conversations between an adult (e.g., a parent or a therapist) and a child, allegedly a victim of sexual abuse, are often admitted, under a variety of hearsay exceptions, to prove that the abuse occurred. (18) These hearsay exceptions, however, operate only one way. What the defendant or his agent said that makes the defendant look guilty is admissible; what they said that makes him look innocent, along with out-of-court statements of disinterested witnesses or third parties, are usually excluded as hearsay. (19)

Despite its frequency, the reliability and credulity of conversational testimony is virtually ignored in scholarly materials. Major textbooks on psychology and law address the role of memory in legal contexts almost exclusively with respect to eyewitnesses, as if this were the only memory relevant in the courtroom. Very little research has addressed conversational testimony per se. (20) Indeed, in contrast to the various safeguards in place for assuring the reliability of other types of evidence there are virtually none for the conversational witness aside from swearing in. For example, an expert's credentials must be assessed and her testimony judged "helpful" to be admissible and even eyewitness accounts are not admissible if unreliable; there is no judicial authority, however, to exclude relevant but unreliable conversational testimony.

Our judicial system rests on a set of assumptions about the ability of witnesses and the capacities of jurors. Our experiences with eyewitness fallibility demonstrate that such assumptions often turn out to be erroneous, resulting in serious injustices. Testimony about conversations that allegedly occurred two, ten or even twenty years ago is commonplace, often constituting the core of the case, and the system assumes that such testimony is either reasonably reliable or its unreliability is adequately discounted by the trier of fact. The primary purpose of this article is to cast doubt on that assumption: conversational witnesses are more common, more likely to be inaccurate, more likely to be believed by jurors, and more likely to produce irreversible errors than eyewitness testimony. Part II highlights the case for further psychological and legal attention to the conversational witness by demonstrating the frequency and finality of conversational testimony. Part III undertakes a review of the existing psychological literature to explore the unique aspects of conversations and the theoretical constructs underpinning the inaccuracy of conversational memory. Part IV undertakes a comparative analysis of eyewitness versus conversational witness performance, with an emphasis on how variables known to affect eyewitness memory may pose even greater challenges to conversational memory. Part V explores the differences in the legal system's response to the dangers inherent in eyewitness versus conversational witness testimony. Finally, in Part VI, we explore a range of reforms that could substantially reduce the risk of erroneous convictions attributable to conversational witnesses.

II. DIMENSIONS OF THE PROBLEM

Eyewitness identification errors, although accounting for 4,000 or more false convictions annually in the United States, (21) are not risk factors in most criminal prosecutions. (22) Conversational testimony, in contrast, appears in most contested criminal prosecutions. The criminal prosecution that does not rely at all upon testimony about what the defendant said or what another person said to the defendant is a rarity.

Statements attributed to the defendant are often the actus reus of the crime, as when the charge is perjury, lying to an official, extortion, an illegal threat, obstruction of justice, bribery, solicitation of unlawful sexual acts, or various kinds of attempts. These statements are often made orally without documentary corroboration. Determining what was actually said by the defendant, as well as the crucial conversational context, frequently rests on the memory and the credibility of a single witness.

Even when the actus reus of the crime includes physical conduct, as it often does even in nonviolent crimes, the prosecution's case commonly rests in part upon testimony about what the defendant said or what was said to him. Conversational testimony helps to reconstruct what the defendant did, the mental state accompanying his conduct, or both. The conversations occur in a myriad of circumstances. Conversational testimony is frequently introduced under the co-conspirator exception to the hearsay rule, which permits a witness to testify about conversations between co-conspirators, even when the defendant was not present. (23) Conversational testimony also comes from cooperating accomplices reporting statements by or to the defendant that evidence a guilty state of mind. In a substantial portion of felony prosecutions, the prosecution obtains such testimony from family members or snitches who quote the defendant as having made incriminating references to his behavior.

The prosecution's case against a particular defendant sometimes rests on a fragment of ambiguous conversation. In United States v. Alvarez, (24) for example, two men told DEA agents of a plan to deliver a load of kitchen appliances to Colombia from a remote airstrip near Miami, Florida, smuggling marijuana on the return flight. Alvarez arrived at the airstrip driving a pickup truck loaded with kitchen appliances, which he then helped load onto the airplane. One of the agents, acting undercover, asked Alvarez if he was going to be at the off-loading site on return. Alvarez "smiled and nodded affirmatively." (25) The court held that this smile and nod, in response to the agent's question (together with the hearsay about the smuggling plan which was not otherwise connected to Alvarez) was sufficient to convict him of conspiring to smuggle marijuana. (26) In another drug case, the evidence connecting Carmine Tramunti to a drug conspiracy was a conversation he had about another's bail. One alleged co-conspirator told Tramunti, "We are having a problem getting Moe Lentini out of prison." Tramunti replied, "Try to get him out," but as for providing collateral for Lentini's bail, "There's nothing I can do about that." (27) The Second Circuit held that this was sufficient to inculpate Tramunti in the conspiracy. (28) In both Alvarez and Tramunti, convictions were based on conversational testimony despite uncertainty about what was actually said, as well as what was intended. In United States v. Quattrone, (29) the key statements made by Quattrone were not in dispute, since he made them in an email, but what he intended by the email was reconstructed from many other communications. Quattrone was accused of obstructing federal investigations by sending an email to his subordinates in an investment banking operation. The National Association of Securities Dealers (NASD), the Securities and Exchange Commission (SEC) and a grand jury had been investigating several initial public offerings and had issued subpoenas. (30) An associate circulated an email to various employees, explaining that under the company's document retention policy no notes, drafts, internal memos or other similar documents were to be retained after the transaction was completed. (31) The email added, "you should catch up on file cleanup." (32) Quattrone then sent an email endorsing the cleanup memo, saying "I strongly advise you to follow these procedures." (33) The Government claimed that this was a corrupt suggestion that documents covered by the subpoenas be destroyed, which Quattrone denied. (34) In this case, conversations around the subject of the email were important in determining what Quattrone intended by his email.

On other occasions, a brief conversation with the defendant may be offered to demonstrate guilty knowledge. In the 2005 trial of Bernard Ebbers, chief executive of bankrupt WorldCom, Ebbers was accused of defrauding investors. Ebbers swore he was unaware of the false accounting used by chief financial officer Scott Sullivan to hide expenses and inflate revenues. (35) Sullivan, however, testified that he told Ebbers that "the only way the company could meet its earnings projections would be to make 'adjustments' to the financial statements." (36) Ebbers replied, according to Sullivan, "We have to hit our numbers." (37) On the basis of such conversational testimony, Ebbers was convicted and sentenced to 25 years in prison. (38)

Conviction or acquittal may turn on the precise wording of out-of-court statements. In the 2001 price fixing trial of A. Alfred Taubman, former head of Sotheby's, his chief accuser, Diana Brooks, testified that Taubman said to her, implying that he wanted her to take the blame for the price-fixing, "You will look good in [prison] stripes." (39) Taubman called to the stand another participant in the conversation, however, who swore that Taubman said, self-mockingly, "Do you think I'd look good in stripes?" (40) This conflicting testimony about Taubman's exact words was a major controversy during Taubman's trial, presumably because Taubman's version of this seemingly unimportant discrepancy, if believed, would discredit the testimony of his accuser. (41) Taubman, however, was convicted. (42)

The statements above went to prove criminal intent. In other cases, the criminal conduct itself is proved by ambiguous statements. In a Texas court, Roy Criner was convicted of rape (the victim was also murdered, but Criner was charged only with the rape) based almost entirely on testimony by his co-workers that Criner claimed he had picked up a hitchhiker and forced her to have oral sex, after which he released her. (43) DNA evidence later proved that Criner was not the rapist. (44) In the highly publicized 2002 trial of Michael Skakel for the 1975 murder of Martha Moxley, Skakel was convicted almost entirely on ambiguous statements he was said to have made on diverse occasions, some of them more than twenty years before the trial. They included, "[I don't] know whether [I] did it," I "may have done it," (45) "I did it" and "I am going to get away with murder. I am a Kennedy." (46)

Conversational testimony is also important when offered to prove that other statements the defendant made were false. In Martha Stewart's prosecution for lying to the SEC and the Federal Bureau of Investigation (FBI), the prosecution claimed that she lied when she said she sold stock without any knowledge that the CEO of the company was selling his stock. (47) Among the witnesses testifying against her was a woman who had been vacationing with her in Mexico when the stock was sold. She testified that Martha told her that the CEO had sold his stock and it is "nice to have brokers who tell you these things." (48)

In many felony cases, the prosecution also depends on testimony by police officers about what the defendant said, either in the form of incriminating admissions or confessions. (49) Even when those statements are reduced to writing and signed by the defendant, their probative value rests substantially on testimony about what the defendant orally said to the police prior to, during and after signing the document. The cogency of such testimony depends not only upon the credibility of the witnesses about what the defendant said but also on the entire context of the interrogation: what was said and done to the defendant by the interrogators.

Since investigations and trials are designed to reconstruct an approximation of past events and the mental states that accompanied those events, it should not be surprising that conversational testimony is present in virtually every trial, criminal or civil. What is less clear is how consequential such testimony is in comparison to that of eyewitness testimony. When the identity of the perpetrator of a crime is at issue and an eyewitness mistakenly identifies the defendant, the false eyewitness testimony is highly likely to have been a causal factor in any jury's decision to convict the accused. The causal role of conversational witnesses is more speculative, since the testimony will often bear not on the identity of the criminal but rather on whether any crime was committed at all. (50) The overall impact of false conversational testimony is still very significant in the criminal process, however, because such testimony is more common, conveys a significant emotional impact, and its impact is more permanent.

Even if the typical conversational witness had less influence upon a jury than an eyewitness, the far greater frequency of conversational witnesses is a counterbalancing factor. To be an eyewitness, a person must have been at the scene of the crime or its immediate aftermath. One who was not seen and identified by others at or near the scene shortly after the crime is unlikely to make a credible eyewitness. Although the testimony of a single eyewitness, like the testimony of a single conversational witness, may suffice to convict without corroboration, (51) there is some inherent corroboration in the testimony of almost any eyewitness. The eyewitness must be in a position, if challenged, to show by evidence other than his own testimony that he was present at a time and place in which he could have observed what he testified about. Consequently, the potential number of eyewitnesses in most prosecutions involving a questioned identity is extremely limited. A conversational witness, however, need only have been in the same geographical area as the defendant during any part of a substantial period of time--often years--and to claim to have conversed with the defendant at some time during that period, or at least to have overheard the defendant conversing with others. Alternatively, the witness could swear to a telephone conversation with the defendant. The number of potential conversational witnesses is, therefore, almost infinite, and their claims to have had a conversation with the defendant or to have overheard a statement by him are extremely difficult to contest.

Conversational witnesses can deliver emotionally powerful testimony that can have persuasive effects far greater than more prosaic albeit more reliable evidence. In the 1989 tax evasion prosecution of Leona Helmsley, for example, despite documents proving that huge personal expenses had been erroneously deducted as business expenses, the most damning evidence against Mrs. Helmsley may well have been the brief testimony of her former maid who quoted her as having said, "We don't pay taxes. Only the little people pay taxes." (52) In the recent trial of Enron executives Kenneth Lay and Jeffrey Skilling for defrauding Enron's investors, the extensive, tedious and complex evidence designed to show that the defendants must have known that the company was on the verge of collapse was bolstered by brief but powerful conversational testimony. Andrew Fastow, former chief financial officer, testified that he told Kenneth Lay that the company was in desperate need of a "massive restructuring" without which it could not long survive. (53) Another former executive, Kevin Hannon, testified to a meeting where some dubious partnerships, designed to hide losses, were discussed. According to Hannon, Jeffrey Skilling said, in apparent reference to the investment community, "They're on to us." (54) In cases like Enron, a few words attributed to the defendant may carry more weight with the jury than a mountain of financial evidence.

Unlike eyewitness testimony, conversational testimony is largely immune from falsification and is therefore more final than is eyewitness testimony. When evidence is developed that an eyewitness identification was erroneous, as when DNA evidence points to a different perpetrator, or the true villain convincingly confesses, the wrongful conviction will often be avoided or, if the evidence is discovered after trial, will be vacated. (55) Cogent evidence rarely exists to establish the falseness of conversational testimony. If the witness had contact with the defendant (or a plausible claim to the same) and no other witnesses were present, the witness's word can rarely be disproven. The witness's motives for testifying can be explored, but this hardly demonstrates that the witness's recollection of the conversation is erroneous. Sometimes the witness may have made inconsistent statements about the conversation that he testified to having had or heard which can be used to undermine his testimony, and sometimes the witness himself may even recant. If the inconsistent statements or recantations are discovered after trial, however, courts are extremely reluctant to vacate convictions based upon newly discovered contradictions or recantations. (56) Their reluctance is based on two assumptions: (1) contradictions and recantations are themselves conversational accounts which have credibility problems: there is little reason to credit them over the sworn testimony given at trial; and (2) even if false, the conversational testimony may have been "harmless", i.e., it may not have caused the guilty verdict. (57) Since jury verdicts are immune from scrutiny into their bases, notions about harmlessness are highly speculative, and courts, eager to uphold the conviction, commonly place a nearly impossible burden on the defendant to prove that the false evidence caused his conviction. (58)

Conversational testimony may be less dramatic than eyewitness identifications, and there is no equivalent to DNA testing to spectacularly prove it wrong. The aggregate and largely irreversible impact of erroneous conversational testimony is no less significant, however, although in the shadowy world of memory, motives, intent and implication, the scale of this impact can be difficult to quantify. In contrast to the highly publicized cases of mistaken identity, the problem of erroneous conversational testimony largely goes unnoticed.

Of course it is not just incriminating conversations which are problematic: testimony about exculpatory conversations may be equally erroneous. Conversations with defendants or third parties might wrongly point to someone else as the perpetrator, such as where a third person is said to have confessed to the crime of which the defendant is accused. However, the risk of erroneous conversational testimony convicting the innocent is much greater than the risk of it setting free the guilty. As noted earlier, our rules of evidence clearly distinguish between out-of-court statements that incriminate the accused and those that exculpate or exonerate him. The latter statements are usually excluded as hearsay. (59)

The plethora of potential conversational witnesses makes them an especially fecund source of witnesses to fill in gaps in the prosecution's case. Convictions based upon conversational testimony are more likely to be final and unimpeachable than is eyewitness testimony. Careful analysis of the existing psychological literature suggests, however, that not only is conversational testimony more common and less reversible than eyewitness testimony, it is also less accurate. It is the reliability of conversational memory to which we now turn.

III. CONVERSATIONAL MEMORY

In contrast to the large body of literature on the factors that influence eyewitness testimony, there are few studies which systematically examine variables that may affect the accuracy of conversational recall. Conversations are poorly understood, and more ambiguous and complex than the relatively simple items subjects are asked to remember--for example, an image, text, or story--in most research in the psychological literature.

Some studies examining memory for oral information have focused on the distortions that result during the process of relaying information from person to person (i.e., the process of rumors). (60) While such material has implications for the accuracy of hearsay evidence, (61) memory for conversations in which one actively participates is a distinct issue to which rumor research does not readily generalize. (62) Other studies examining verbal memory have focused on words and sentences, which, though relevant, are not fully adequate for understanding how larger and much more complex conversations are remembered. (63) Similarly, memory for a transcript of a conversation may or may not be readily generalized to memory for a live, interactive conversation. (64)

The dearth of studies examining "conversational memory" in the true sense of the word, as opposed to memory for any oral material or conversation transcripts, has probably contributed to the continuing complacent acceptance of the general accuracy of conversational testimony. Nevertheless, even the limited studies available are sufficient to support the following generalizations about conversational memory, each of which will be examined below: (A) conversational memory is astoundingly poor; (B) people remember gist information better than surface information; (C) conversational memory is strongly influenced by motivational biases; (D) conversational memory is subject to schema-driven or contextual errors; (E) conversational memory is extremely malleable; (F) neither witness consistency nor witness confidence implies accuracy of conversational memory; and (H) the conversation as the object to be remembered is unique in that it is cross-modal. Before addressing these propositions individually, it is necessary to understand the dominant psychological model for conversational memory, fuzzy trace theory.

Fuzzy trace theory proposed a psychological model--subsequently well validated--consisting of two parallel memories for conversations. The first, surface memory, functions like a videotape, including the actual words, phrases, intonation and gestures used in the conversation. The second, gist memory, is a construct created by the listener from the underlying meaning of what is communicated, as the listener understands it. (65) Thus, a given speech act, such as a request to have the door closed (gist), may be expressed in multiple forms (surface). The surface form, in turn, can vary in syntax [imperative ("close the door!"), interrogative ("could you close the door?"), or declarative ("it would be nice if you closed the door")]; in degree of directness [direct ("could you close the door"?) or indirect ("it's getting chilly in here")]; and in prepositional content ["I need the door closed" versus "I would like the door to be closed"].

The two memories differ in terms of their duration (rate of decay) and accuracy (the ability to resist false identifications and retroactive interference (66)). Surface memory decays rapidly and resists false identifications, but is subject to retroactive interference. Gist memory is longer-lasting and subject to misidentification, but resists retroactive interference. (67) Further research has even allocated gist and surface memory to different anatomical regions of the brain, namely the frontal...

NOTE: All illustrations and photos have been removed from this article.



More articles from American Criminal Law Review
Corporations cry uncle and their employees cry foul: rethinking prosec..., January 01, 2007
Crime and (with a lag) punishment: the implications of discounting for..., January 01, 2007
Satisfy the demands of justice: embrace electronic recording of custod..., January 01, 2007
Children as adults: the transfer of juveniles to adult courts and the ..., January 01, 2007

Looking for additional articles?
Search our database of over 3 million articles.

Looking for more in-depth information on this industry?
Search our complete database of Industry & Market reports by text, subject, publication name or publication date.

About Goliath
Whether you're looking for sales prospects, competitive information, company analysis or best practices in managing your organization, Goliath can help you meet your business needs.

Our extensive business information databases empower business professionals with both the breadth and depth of credible, authoritative information they need to support their business goals. Whether it be strategic planning, sales prospecting, company research or defining management best practices - Goliath is your leading source for accurate information.