Dr. Papcun's Report on the Rodney King trial appears in the book "Letters of the Century"
(Read the letter)
by Grunwald and Adler
"Your appearance as the final witness in the Rodney King trial was decisive as to the outcome."---Milton Grimes, Esq.

"Dr. Papcun is a star!"---Attorney subsequent to Dr. Papcun's deposition, following which the opposition declined to proceed

"Please send your bill in before you find out how delighted the client is." --Attorney following a directed verdict subsequent to Dr. Papcun's testimony

Voice Lineups


Voice lineups defined  |  The problem  |  Testing a lineup  |  Case Study
Guidelines for voice lineups  |  Analysis and Critique of procedures  |  Graphics


Voice lineups defined

A voice lineup is a test in which a listener hears several voices and tries to tell whether a voice she has heard previously is among them, and if so, which it was. Conducted properly, a voice lineup can protect the innocent. On the other hand, conducted improperly, a voice lineup can effectively point an accusing finger at an innocent person.

The problem

Your speech says so much about you that it is very difficult to create a proper voice lineup. Your speech can reveal your age, your health, your level of education, your regional dialect, and many other factors. Even the location in which a recording is made can reveal information. For example, a recording made in a jail cell will sound different from a recording made in a book-lined lawyer's office, so if one of the subjects is recorded in a jail cell and the others are recorded in their offices, the lineup effectively points out the person who was recorded in the jail cell as being a suspect. A detailed discussion of how to properly construct and administer a voice lineup is presented below in Guidelines for voice lineups.
( return to top )

Testing a lineup:

Whether a voice lineup is biased can be tested with a surrogate jury, in which a group of listeners who have no connection with the case are asked questions about the people in the lineup. If they can discriminate the accused from the others, it can be argued that the purported identification by the witness is meaningless. Worse, the voice lineup has actually served to "finger" the accused.(return to top)

Case Study - A Voice Lineup for Donald Duck:

As he was entering the Home Depot by way of the Garden Center, the man in the long white paper coveralls asked the employee at the cash register (Lucas Vocos) where he could locate a manager. Possibly because the man was wearing a face mask of the kind that painters wear to protect themselves from spatters and fumes, the employee did not hear him the first time so the man repeated the question. The employee pointed out the direction of the manager’s office.

Because the man was dressed in white coveralls and had on a bright yellow hard hat and gloves, another employee in the Garden Center (Dennis Malone) quipped to him “You know, you look like Donald Duck” but the man was in no mood for humor and did not respond. The employee noticed that the man’s face mask was cut and advised “Well, you need a new mask. They’re over in hardware.” However, the man did not go over to the hardware section. Instead, he stood around, shuffling about in the Garden Center. The employee inquired if he needed any help. The man answered “No, I’m okay” and kept fooling with his mask. The employee persisted and said “Well, are you sure? You look like, you know, you need a manager.” The man looked around and said “Well, do you see one?” The employee said “No, I don’t see one, but I can get one for you.” Again, the man said “No, I’m okay,” so the employee walked away.

The manager on duty was Tom Egan, ex-Marine, married father of three-year-old twin girls, well-respected by other store employees. The man in the white coveralls approached Egan, pulled a gun out of a black bag and demanded money from the safe. As he pulled out the gun, there also fell from the bag a dirty old sock. Egan told the man he did not have access to the safe. As the man walked away, Egan followed him and pleaded with him to leave and not harm anyone, whereupon the man turned on Egan and shot him in the stomach.

The man stepped over Egan’s dying body and went to the Pro Desk. To the employee at the Pro Desk he said “Give me all the money.” After the employee handed him the first cash drawer, the man told him “There’s more. There’s more. Get all the stuff underneath. You have more.” The man in the white painter’s coveralls took off with about $500 in cash.

A customer who was parked next to the robber’s vehicle saw him leave the store, get into a silver Ford Windstar van and drive away. Police surveillance subsequently spotted the van, which was owned by a relative of Jason Russell Richardson, a thirty-seven year old ex-convict who was then on parole and living with his mother in the nearby town of Oceanside, California. The hunt for Richardson was featured on the television show “America’s Most Wanted.” Richardson had spent much of his adult life in jail on convictions including robbery, burglary, spousal abuse, molesting a minor and rape. Evidently, Richardson did not see the show because he reported to his parole officer on schedule. When Richardson reported to his parole officer, he was taken into custody.

Egan was taken by ambulance to a local hospital. He died about two hours later. Richardson was accused of murder with special circumstances, meriting the death penalty.

In order to demonstrate that Richardson was in fact the man in the white coveralls, Detective Mike Van Cleve of the Tustin police Department prepared both a visual lineup and a voice lineup, each consisting of the suspect Jason Russell Richardson and five “fillers,” i.e., people such as police officers who were not suspects. The visual lineup was administered to store employees who had had seen the killer and to customers who had seen him in the parking lot.

The results of the visual lineup were as follows: The employee who saw the man in the white coveralls enter the store (Lucas Vocos) said “If it’s anyone, it’s this guy” (Richardson). Similarly, the other employee from the Garden Shop area (Dennis Malone) said “It looks more like three (Richardson) than any of the others.”The employee at the Pro Desk (Francisco Cerda) selected Richardson on the basis of “the shape of his head and his jaw size.” Cerda was 100% certain of his choice. An employee who was in the receiving area (Maria Guerrero) covered up the forehead and mouth of each of the photos, and selected Richardson on the basis of the eyes. Another employee who had been in the receiving area (Isiah Atuatasi )selected the photo of Richardson, but said he did so from having seen Richardson on television after his arrest. The customer who saw the man in the coveralls run from the store and get into the Ford Windstar chose Richardson on the basis of the shape of his head and his skin color. Eleven observers said they were unable to identify the killer from the visual lineup.

A voice lineup is active in the sense that it requires that the subject do something; namely, speak, which by its nature is likely to reveal information about his educational level , his ethnic background, and even certain aspects of his mood or personality, and other characteristics. By contrast, a visual lineup is passive in the sense that the police can take a picture of the subject and show it, along with pictures of others, to the witnesses, who may then decide whether they have previously seen one of the people in the pictures. The subject does not have to do anything besides allow a picture to be taken. Detective Mike Van Cleve of the Tustin Police Department constructed a voice lineup that consisted of Richardson’s voice plus the voices of five others including police detectives and an IT specialist who worked for the department. Each person spoke the words “She works for an environmental agency called Three E Companies,” which Richardson had spoken in the course of his interview when he was first taken into custody.

The results of the voice lineup were as follows: Detective Van Cleve played the voice lineup three times for the employee at the Garden Center cash register (Lucas Vocos), whom the man had twice asked where he could locate a manager. Vocos said he was unable to positively identify anyone in the lineup.

The Garden Center employee who had quipped that the man in the white coveralls and the yellow helmet looked like Donald Duck (Dennis Malone) said he thought the voice was the #3 voice (which was NOT Richardson’s voice). Detective Van Cleve then told Malone he wanted him to be 100% certain before choosing a voice and Malone said “I’m pretty sure it’s number three.” Van Cleve asked how certain he was and he said “Probably ninety eight (percent).”

Employee Isiah Atuatasi, who was in the receiving area, said the voice of the killer was the #4 or the #6 voice, but he couldn’t be sure.

In her introduction to Eyewitness Evidence, A Guide for Law Enforcement1, Attorney General Janet Reno states

Eyewitnesses frequently play a vital role in uncovering the truth about a crime. The evidence they provide can be critical in identifying, charging, and ultimately convicting suspected criminals. That is why it is absolutely essential that eyewitness evidence be accurate and
reliable. One way of ensuring we, as investigators, obtain the most accurate and reliable evidence from eyewitnesses is to follow sound protocols in our investigations.

True as those observations are for eyewitness evidence, they are equally important for evidence based on hearing, which we may think of as “earwitness” evidence. With the ubiquity of recording media, such evidence is becoming increasingly common. It is important, therefore, that we examine the issues that present themselves with regard to evidence based on earwitness evidence, both in terms of similarities to and differences from eyewitness evidence.

Working with Richardson’s attorney Julie Anne Swain, who was Associate Defender for the Orange County, California Homicide Unit, I prepared detailed guidelines for voice lineups based closely on federal guidelines for visual lineups, though taking into account the differences required by the differences between visual and audio media. The treatise is below under Voice Lineup Guidelines.

When I compared the voice lineup in this case to the requirements set forth in my treatise, I found numerous respects in which it failed to meet the proper standards. The simple fact that Richardson had spoken the words in normal conversation, whereas the others had not, caused a noticeable difference in speech style. Moreover, Richardson’s dialect differed substantially from that of the other speakers, who were relatively uniform among themselves. Furthermore, the lineup was presented improperly in many respects such as that Detective Van Cleve appears to have tried to influence the certainty judgment of a witness. In support of my findings, I prepared graphical analyses such as pitch tracks of the speakers’ voices, acoustic spectrograms, and phonetic transcriptions of the dialects.

The report is below, under the heading Richardson Lineup Report.

When presented with my deposition detailing my findings, the prosecutor decided not to use the voice lineup evidence. I have to say that I do not entirely fault Detective Van Cleve for all of the errors in the construction and presentation of the voice lineup. Because a valid voice lineup requires the active participation of the subjects, it is much more difficult to prepare and administer than a valid visual lineup. I hope that my treatise on this topic will help.

There was another piece of evidence, however, that was unknown to me at the time of the time of my deposition. When the police swept the store, they found the dirty sock that had fallen out of the bag that contained the gun. The sock contained five rounds of Winchester .38 Special ammunition, which was the kind of ammunition that killed Egan. From the sock, the police were able to collect DNA evidence. Recall that Richardson had previously been convicted of rape. The DNA from the sock matched Richardson’s prior DNA evidence as well as Richardson’s contemporary DNA.

The story does not end there. The sock also contained DNA from Richardson’s wife, his mother’s boyfriend, and Richardson’s half brother, who looks very much like him.

Do the judgments revealed by the lineups constitute evidence beyond a reasonable doubt? That is not for me to say, but I think it is at least arguable that they do not. Nonetheless, but for a sock dropped in a moment of haste, the probability of Richardson’s conviction of capital murder with special circumstances and a consequent death sentence would have been far less likely.

Epilogue: Richardson was found guilty.
( return to top )


Guidelines for Voice Lineups

George Papcun, Ph.D.
March, 2010

Outline

Introduction


The importance of proper procedures for “earwitness” testimony Guidelines based on DOJ recommendations for visual lineups Differences between visual and audio media Recordings Requests to hear a voice again

A. Composing Voice Lineups B. Instructing the Witness Prior to Hearing a Voice Lineup C. Conducting the Identification Procedure D. Recording Identification Results Summary

References

Appendix: Research Issues

Introduction

In her introduction to Eyewitness Evidence, A Guide for Law Enforcement, 1 Attorney General Janet Reno states "Eyewitnesses frequently play a vital role in uncovering the truth about a crime. The evidence they provide can be critical in identifying, charging, and ultimately convicting suspected criminals. That is why it is absolutely essential that eyewitness evidence be accurate and reliable. One way of ensuring we, as investigators, obtain the most accurate and reliable evidence from eyewitnesses is to follow sound protocols in our investigations."

True as those observations are for eyewitness evidence, they are equally important for evidence based on hearing, which we may think of as “earwitness” evidence. With the ubiquity of recording media, such evidence is becoming increasingly common. It is important, therefore, that we examine the issues that present themselves with regard to evidence based on earwitness evidence, both in terms of similarities to and differences from eyewitness evidence. This treatise is particularly devoted to voice lineups.

The guidelines set forth in this document are based on DOJ recommendations for visual lineups 1,2. However, it must be recognized that there are certain inherent differences between voice lineups and visual lineups. Voice is actively produced by the subject, whereas appearance is passive. Voice may convey certain information not necessarily conveyed by appearance, such as educational level, social and ethnic background. It is important, therefore, that the suspect and the fillers be matched on these considerations. Speaker identity is conveyed by details of pronunciation as well as by the anatomy of the speaker. It is appropriate, therefore, to transcribe details of the pronunciation with the detailed phonetic transcription to illustrate the differences and similarities of pronunciation among the several speakers. Recordings of the speakers in the lineup can be played along with the phonetic transcription in order to illustrate details of pronunciation.

The differences between audio and visual media impose choices regarding how exemplars are to be obtained. What should the subjects say? How long should the utterance be? Can it be obtained from the natural speech of the subject? Should the utterance be obtained by having the subjects read the words? Should the utterance be obtained by having the subjects repeat them following a speaker who provides an example? Should all the subjects say the same things? These issues are addressed below in Section A. Whatever decisions are made with respect to these issues, it is imperative that they be applied to the suspect and to the fillers (i.e., nonsuspects) uniformly.

Recordings are imperfect. They may fail to record and play back all the voice frequencies or may limit the dynamic range of the voice. The effect of an imperfect recording may be compared to using a photograph that is blurred or insufficiently detailed. These effects may have the consequences of failing to provide sufficient information to allow an identification or of limiting information that would exclude a speaker.

With photographs, it is usually self-evident whether a photograph depicts a subject clearly enough for identification or elimination. It may be more difficult to determine whether a recording portrays a subject’s voice adequately for identification or elimination. Moreover, a recording can distort a voice; for example, by making a voice sound deeper or higher-pitched than the actual voice. Because recordings are imperfect, it may be advantageous to conduct a live voice lineup, though at the possible cost of other complications. These issues are addressed below in Section A.

Voice is necessarily presented sequentially, whereas visual identification may be made from simultaneously presented materials. Moreover, because of the inherently sequential and hence evanescent nature of audio material, listeners frequently request that voices be repeated. The implications of these issues are addressed below in Section B.

It is vital that lineups are presented in such a way that the investigator is not conveying information about which of the voices is the suspect. This issue is addressed below in Section C.

A. Composing Voice Lineups
Principle: Fair composition of a voice lineup enables the witness to provide a more accurate identification or nonidentification.

Policy: The investigator shall compose the voice lineup in such a manner that the suspect does not unduly stand out. Additionally, the investigator shall compose a voice lineup with material appropriate and sufficient to provide an adequate basis for identification or nonidentification.

Discussion:

Selecting the utterance material
As mentioned in the introduction, the selection of material to be spoken is vitally important for conducting a valid voice lineup. The spoken material should resemble as closely as possible the material spoken by the perpetrator at the time the voice was heard originally. Ideally, the material for the lineup should consist of the same utterance as was spoken by the perpetrator at the time it was heard originally, spoken in an natural way, though this may not necessarily be available. Typically, it is necessary to rely on the memory of the witness to determine what the original utterance was. Within practical limits, the material used in the lineup should be as long as the utterance originally spoken. Whatever approach is chosen for selecting the utterance material, it is imperative that the same approach be used for all of the speakers, suspect and fillers alike.

Producing the utterance material
If the subject is well educated and reads fluently, it may be appropriate to have him (and the fillers, assuming they also are well educated) read the materials. An alternative approach is to have the investigator read the utterance phrase by phrase, and have the suspect repeat the phrases onto the recording. It is may be possible to capture natural speech by asking the suspect and the fillers the same meaningful questions, though their answers may not all be the same. Again, whatever approach is used for producing the utterance material, it is imperative that the same approach be used for all of the speakers. For example, it is inappropriate to record natural speech from the suspect and have the fillers say the same phrase from memory of his utterance or by reading it. Such differences in mode of production produce differences in style, fluency and elision effects that may identify the suspect as distinct from the fillers.

Uniformity of accent, dialect, educational level, and speech characteristics
If the subject has an accent influenced by a foreign language or a distinct regional or social dialect, the fillers should have similar characteristics. For example, It would be inappropriate to present a suspect who has a lisp along with fillers who do not lisp. Similar considerations apply to the influence of social dialects, foreign languages and general voice characteristics. It would be inappropriate to present a suspect who speaks in a black urban dialect along with fillers who do not speak in that dialect. It would be inappropriate to present a suspect who has a Spanish accent along with fillers that do not have a Spanish accent. It would be inappropriate to present a subject with a very deep voice along with fillers with high-pitched voices. The suspect and the fillers should have similar levels of fluency in English (or whatever language is being used for the lineup).

Recording and playing back the lineup
The recording and playback equipment should be capable of recording and playing back the main range of the human voice faithfully and in an undistorted manner. At a minimum, this requires that the system record and playback frequencies from 120 Hz. to 5,500 Hz. with an amplitude deviation of no more than plus or minus 6 dB. These measurements can be made by a qualified expert.

Recorded versus live lineups
As discussed above, recordings are imperfect. Therefore, it may be advantageous to conduct a live voice lineup. This can be done by having the suspect and the fillers stand behind the witness, behind a screen or in an adjacent room. If during the course of a live voice lineup the witness sees the suspect or the fillers, the basis for the identification is confounded between audio and visual media. Because the suspect and the fillers must perform for a live voice lineup, the situation may be less controlled than for a recorded lineup and less convenient.

Procedure for Composing a Recorded Lineup:

In composing a recorded lineup, the investigator should:
  1. Include only one suspect in each identification procedure.
  2. Select fillers who generally fit the witness’ description of the voice of the perpetrator. When there is a limited/inadequate description provided by the witness of the perpetrator’s voice, or when the witness’ description of the perpetrator’s voice differs significantly from the voice of the suspect, fillers should resemble the voice of the suspect in significant characteristics, particularly those set forth in the discussion above.
  3. If multiple recordings of the suspect are reasonably available to the investigator, select a recording that resembles the witnesses’ description of the suspect’s voice at the time of the incident.
  4. Include a minimum of five fillers per identification procedure.
  5. Consider that complete uniformity of features is not required. Avoid using fillers whose voices so closely resemble the suspect’s voice that a person familiar with the suspect might find it difficult to distinguish the suspect from the fillers.
  6. Create consistency between the suspect and fillers with respect to any distinctive characteristics (e.g., dialect, foreign accent, speech impediment or other issues set forth in the discussion above).
  7. If more than one lineup is to be conducted, consider placing suspects in different positions in each lineup, both across cases and with multiple witnesses in the same case. Position the suspect randomly in the lineup.
  8. When playing the voice of a new suspect, avoid reusing fillers in lineups played to the same witness.
  9. Ensure that no writings or information concerning previous arrest(s) will be visible to the witness.
  10. Listen to the spread, once completed, to ensure that the suspect does not unduly stand out.
  11. Preserve the presentation order of the recorded lineup. In addition, the recordings themselves should be preserved in their original condition.

Procedure for Composing a Live Voice Lineup:

In composing a live voice lineup, the investigator should:
  1. Include only one suspect in each identification procedure.
  2. Select fillers who generally fit the witness’ description of the voice of the perpetrator. When there is a limited/inadequate description of the perpetrator’s voice provided by the witness, or when the description of the perpetrator’s voice differs significantly from the voice of the suspect, fillers should resemble the suspect in significant voice characteristics.
  3. Consider placing suspects in different positions in each lineup, both across cases and with multiple witnesses in the same case. Position the suspect randomly unless, where local practice allows, the suspect or the suspect’s attorney requests a particular position.
  4. Include a minimum of four fillers (nonsuspects) per identification procedure.
  5. When showing a new suspect, avoid reusing fillers in lineups played for the same witness.
  6. Consider that complete uniformity of voice characteristics is not required. Avoid using fillers who so closely resemble the suspect that a person familiar with the voice of the suspect might find it difficult to distinguish the suspect from the fillers.
  7. Create consistency between the suspect and fillers with respect to any distinctive characteristics (e.g., dialect, foreign accent, speech impediment or other issues set forth in the discussion above).
Summary:
The above procedures will result in a voice lineup in which the suspect does not unduly stand out. An identification obtained through a voice lineup composed in this manner may have stronger evidentiary value than one obtained without these procedures.

B. Instructing the Witness Prior to Hearing a Voice Lineup
Principle:
Instructions given to the witness prior to viewing a voice lineup can facilitate an identification or nonidentification based on his/her own memory.

Policy:
Prior to presenting a lineup, the investigator shall provide instructions to the witness to ensure the witness understands that the purpose of the identification procedure is to exculpate the innocent as well as to identify the actual perpetrator.

Discussion:
Requests to repeat a voice
If a witness asks to hear a voice again, it may be played or spoken as requested. However, the witness should not be allowed to hear another voice before rendering an opinion on the voice that was just heard. The reasoning underlying this constraint is that identification or elimination is to be made with regard to an absolute standard rather than by relative comparison of two or more voices. The issue is not whether one or another voice sounds more like that of the perpetrator, but whether the listener believes the voice is in fact the voice of the perpetrator.

Procedure for Instructing the Witness Prior to Hearing a Recorded Voice Lineup

Prior to presenting a recorded voice lineup, the investigator should:
  1. Instruct the witness that he/she will be asked to listen to a set of recordings.
  2. Instruct the witness that it is just as important to clear innocent persons from suspicion as to identify guilty parties.
  3. Instruct the witness that individuals played in the voice lineup recordings may not sound exactly as they did on the date of the incident because voice is subject to variation depending on situations such as whether the speaker had a cold or was under stress.
  4. Assure the witness that regardless of whether an identification is made, the police will continue to investigate the incident.
  5. Instruct the witness that the procedure requires the investigator to ask the witness to state, in his/her own words, how certain he/she is of any identification.
Procedure for Instructing the Witness Prior to Hearing a Live Voice Lineup

Prior to presenting a live voice lineup, the investigator should:
  1. Instruct the witness that he/she will be asked to listen to a group of individuals.
  2. Instruct the witness that it is just as important to clear innocent persons from suspicion as to identify guilty parties.
  3. Instruct the witness that individuals present in the lineup may not sound exactly as they did on the date of the incident because voice characteristics may vary depending on whether the subject had a cold or was under stress.
  4. Instruct the witness that the person who committed the crime may or may not be present in the group of individuals.
  5. Assure the witness that regardless of whether an identification is made, the police will continue to investigate the incident.
  6. Instruct the witness that the procedure requires the investigator to ask the witness to state, in his/her own words, how certain he/she is of any identification.
  7. Create consistency between the suspect and fillers with respect to any distinctive characteristics (e.g., dialect, foreign accent, speech impediment or other issues set forth in the discussion above).
Summary:
Instructions provided to the witness prior to presentation of a lineup will likely improve the accuracy and reliability of any identification obtained from the witness and can facilitate the elimination of innocent parties from the investigation.

C. Conducting the Identification Procedure
Principle:
The identification procedure should be conducted in a manner that promotes the reliability, fairness, and objectivity of the witness’ identification.

Policy:
The investigator shall conduct the lineup in a manner conducive to obtaining accurate identification or nonidentification decisions.

Discussion:
Requests to repeat a voice
If a witness asks to hear a voice again, it may be played or spoken as requested. However, the witness should not be allowed to hear another voice before rendering an opinion on the voice that was just heard. The reasoning underlying this constraint is that identification or elimination is to be made with regard to an absolute standard rather than by relative comparison of two or more voices. The issue is not whether one or another voice sounds more like that of the perpetrator, but whether the listener believes the voice is in fact the voice of the perpetrator.

Statement of certainty
It is standard in scientific research to constrain the listeners’ responses about certainty to some well-defined scale such as the following: “On a scale of one to ten, where ten means ‘absolutely certain’ and one means ‘completely uncertain,’ how sure are you that this is the person you heard before?” For scientific applications, constraining responses in this way has the advantage of making responses comparable to one another across tasks and across people. By contrast, legal usage recommends that no such scale be imposed. Rather, it is proper for witness to express certainty in his/her own words. This is proper because it is a way of revealing how the witness thinks about the identification and his/her certainty.

Confirmation as to suspect choice
If the witness states that the voice of the suspect is the voice heard previously, it is not proper to confirm that judgment. Providing that kind of information to the witness is improper because there is always the possibility (at least until trail and conviction, and possibly thereafter) that some other suspect will be developed. Providing such information irretrievably compromises the witness for future independent identifications.

Stating what other witnesses have said It is improper to state to a witness what other witnesses have said, either by way of selecting or eliminating a voice.

Avoiding cues to the witness
It is important that the person who is administering the lineup use care not to give cues to the witness as to which of the voices is that of the suspect. This issue has an interesting history beginning with a horse named Hans. In Germany in the 1890s. a performing horse named “Clever Hans” (der Kluge Hans, in German) tapped out the answers to simple arithmetic questions that were posed to him by his owner and trainer, Wilhelm von Osten, who was a retired mathematics teacher. Von Osten would ask questions like “Hans, what is the sum of two and three?” and Hans would tap his hoof five times. Especially in light of Darwin’s then recent theory of evolution, animal intelligence was a topic of great interest at the time. Hans’ abilities attracted such wide attention that the German board of education appointed a commission to investigate them. The commission found that Hans could get the right answers even when questioners other than von Osten posed the questions, and even when von Osten was not present, thus precluding the possibility that von Osten was secretly giving the answers to Hans. In 1904 the commission issued a report that concluded that Hans’ abilities were real. Yet one member of the commission, a psychologist named Oskar Stumpf, continued the investigation. He found that when Hans wore blinders and therefore could not see the questioner, Hans did not get the answer right and sometimes became so ornery that he bit the questioner. Closely studying the situation, Stumpf observed that questioners usually reacted with subtle differences in posture and gaze when Hans reached the right answer. Hans was extraordinarily sensitive to such changes, and used them to get the right answer. Typically, the questioners themselves were unaware of their cues to Hans. In psychology, the ability of those who are administering tests or experiments to subtly, and even unknowingly, convey information about the answers to subjects is known in psychology as the “Clever Hans Effect.”

The question then arises as to whether investigators may give such subtle cues to witnesses, albeit unknowingly, and whether witnesses are as sensitive to them as Clever Hans. At least, the investigator should be very careful not to provide subtle cues. A deeper solution to this problem would be for lineups to be administered by investigators who do not know which of the exemplars is the suspect. A comprehensive analysis of the Clever Hans history, along with experiments that illustrate its applicability to human subjects is available in Rosenthal, Robert, 1998.

Procedure for Conducting the Identification Procedure for a Recorded Voice Lineup

When presenting a recorded voice lineup, the investigator should:
  1. Provide listening instructions to the witness as outlined in Subsection B, “Instructing the Witness Prior to Listening to a Lineup.”
  2. Provide the following additional listening instructions to the witness:
    a. Individual recordings will be heard one at a time.
    b. The recordings are in random order.
    c. Take as much time as needed in making a decision about each
    voice before moving to the next one.
    d. All voices will be played, even if an identification is made; or the procedure will be stopped at the point of an identification (consistent with jurisdictional/departmental procedures).
  3. Confirm that the witness understands the nature of the sequential procedure.
  4. Present each recording to the witness separately, in a previously determined order.
  5. Avoid saying anything to the witness or otherwise doing anything that may influence the witness’ selection.
  6. Ask the witness to state, in his own terms, how certain he/she is of the identification.
  7. If an identification is made, avoid reporting to the witness any information regarding the individual he/she has selected prior to obtaining the witness’ statement of certainty.
  8. Record any identification results and witness’ statement of certainty as outlined in subsection D, Recording Identification Results.”
  9. Document in writing the voice lineup procedures, including:
    a. Identification information and sources of all recordings used.
    b. Names of all persons present at the voice lineup
    c. Date and time of the identification procedure
  10. Instruct the witness not to discuss the identification procedure or its results with other witnesses involved in the case and discourage contact with the media.
Procedure for Conducting the Identification Procedure for a Live Voice Lineup:

When presenting a live voice lineup, the investigator/lineup administrator should:
  1. Provide listening instructions to the witness as outlined in subsection B “Instructing the Witness Prior to Hearing a Lineup.”
  2. Provide the following additional viewing instructions to the witness:
    a. Voices will be heard one at a time.
    b. The voices will be presented in random order.
    c. The witness is to take as much time as needed in making a decision about each voice before moving to the next one.
    d. If the person who committed the crime is present, the witness is to identify him/her.
    e. All voices will be presented, even if an identification is made; or the procedure will be stopped at the point of an identification (consistent with jurisdictional/departmental procedures).
  3. Instruct all those present at the lineup not to suggest in any way the position or identity of the suspect in the lineup.
  4. Present each voice to the witness separately, in a previously determined order, removing those previously heard.
  5. Ensure that any identification actions (e.g., conversing, moving) are performed by all members of the lineup.
  6. Avoid saying anything to the witness or otherwise doing anything that may influence the witness’ selection.
  7. Ask the witness to state, in his own terms, how certain he/she is of the identification.
  8. If an identification is made, avoid reporting to the witness any information regarding the individual he/she has selected prior to obtaining the witness’ statement of certainty.
  9. Record any identification results and witness’ statement of certainty as outlined in subsection D, “Recording Identification Results.”
  10. Document the lineup procedures and content in writing, including:
    a. Identification information of lineup participants.
    b. Names of all persons present at the lineup.
    c. Date and time the identification procedure was conducted.
  11. Document the lineup by video. This documentation should be of a quality that represents the lineup clearly and fairly.
  12. Instruct the witness not to discuss the identification procedure or its results with other witnesses involved in the case and discourage contact with the media.
Summary:
The manner in which an identification procedure is conducted can affect the reliability, fairness, and objectivity of the identification. Use of the above procedures can minimize the effect of external influences on a witness’ memory.

D. Recording Identification Results

Principle:
The record of the outcome of the identification procedure accurately and completely reflects the identification results obtained from the witness.

Policy:
When conducting an identification procedure, the investigator shall preserve the outcome of the procedure by documenting any identification or nonidentification results obtained from the witness.

Discussion:
Contradictory or otherwise anomalous responses
Witnesses may occasionally give contradictory or otherwise anomalous responses. For example, they may identify more than one of the voices as the voice they heard previously, even though the investigator knows that all the voices in the lineup are different from each other. Witnesses may say that one voice is more like the voice heard previously than some other voice (even though they have been informed clearly that such a judgment is not the goal of an identification exercise). It is important to record such responses, as they may yield useful information about the witness as well as information about the voices.

Descriptions of the voice
Witnesses sometimes provide descriptions of a voice they heard previously. Such descriptions should be noted whether or not the witness is able to make an identification.

Procedure:
When conducting an identification procedure, the investigator should:
  1. Record both identification and nonidentification results in writing, including the witness’ own words regarding how sure he/she is.
  2. Ensure results are signed and dated by the witness.
  3. Record even results that are paradoxical such as when the witness may identify more than one of the voices in the lineup as the voice that was heard previously.
  4. Ensure that no materials indicating previous identification results are visible to the witness.
  5. Ensure that the witness does not write on or mark any materials that will be used in other identification procedures.
Summary:
Preparing a complete and accurate record of the outcome of the identification procedure improves the strength and credibility of the identification or nonidentification results obtained from the witness. This record can be a critical document in the investigation and any subsequent court proceedings.
( return to top )

References
Citation: 1. Eyewitness Evidence:
A Guide for Law Enforcement
U.S. Department of Justice
Office of Justice Programs
National Institute of Justice
Technical Working Group for Eyewitness Evidence
October 1999
Washington, DC 20531

Citation 2. Eyewitness Evidence:
A Trainer’s Manual for Law Enforcement
U.S. Department of Justice
Office of Justice Programs
National Institute of Justice
September 2003
Washington, DC 20531
""CLEVER HANS" AGAIN.; Expert Commission Decides That the Horse Actually Reasons.". The New York Times. 1904-10-02.

Appendix: Research Issues
Scientific research indicates that identification procedures such as lineups and photo arrays produce more reliable evidence when the individual lineup members or recordings are played for the witness sequentially— one at a time—rather than simultaneously. Although some police agencies
currently use sequential methods of presentation, there is not a consensus on any particular method or methods of sequential presentation that can be recommended as a preferred procedure; although sequential procedures are included in this guide, it does not indicate a preference for
sequential procedures.

It may happen that a subject is disguised, masked, camouflaged or otherwise partly obscured, and speaks to the witness. From some combination of cues, the witness identifies someone. It may be of interest to know which cues were relevant to the detection. Presumably, the investigating officer can ask the witness about this matter. Nonetheless, it is not clear how to validate the answer. In practical terms, the best procedure is to conduct a separate voice lineup and visual lineup. Otherwise the cues are irretrievably confounded.

It is standard in scientific research to constrain the listeners’ responses about certainty to some well-defined scale such as the following: “On a scale of one to ten, where ten means ‘absolutely certain’ and one means ‘completely uncertain,’ how sure are you that this is the person you heard before?” For scientific applications, constraining responses in this way has the advantage of making responses comparable to one another across tasks and across people. By contrast, legal usage recommends that no such scale be imposed. Rather, it is proper for witness to express certainty in his/her own words. This is proper because it is a way of revealing how the witness thinks about the identification and his/her certainty.

It is common practice in visual identification procedures to ask the witness “What it was about the person’s face or body that made you pick this person or eliminate this person?” Witnesses can frequently provide cogent answers to such questions, and useful information about the suspect. Describing comparable characteristics of voices is more difficult for most people. In such cases, an expert in phonetics can help provide appropriate ways of describing voices and ask the witness about them.

It is important that the person who is administering the lineup use care not to give cues to the witness as to which of the voices is that of the suspect. This issue has an interesting history beginning with a horse named Hans. In Germany in the 1890s. a performing horse named “Clever Hans” (der Kluge Hans, in German) tapped out the answers to simple arithmetic questions that were posed to him by his owner and trainer, Wilhelm von Osten, who was a retired mathematics teacher. Von Osten would ask questions like “Hans, what is the sum of two and three?” and Hans would tap his hoof five times. Especially in light of Darwin’s then recent theory of evolution, animal intelligence was a topic of great interest at the time. Hans’ abilities attracted such wide attention that the German board of education appointed a commission to investigate them. The commission found that Hans could get the right answers even when questioners other than von Osten posed the questions, and even when von Osten was not present, thus precluding the possibility that von Osten was secretly giving the answers to Hans. In 1904 the commission issued a report that concluded that Hans’ abilities were real. Yet one member of the commission, a psychologist named Oskar Stumpf, continued the investigation. He found that when Hans wore blinders and therefore could not see the questioner, Hans did not get the answer right and sometimes became so ornery that he bit the questioner. Closely studying the situation, Stumpf observed that questioners usually reacted with subtle differences in posture and gaze when Hans reached the right answer. Hans was extraordinarily sensitive to such changes, and used them to get the right answer. Typically, the questioners themselves were unaware of their cues to Hans. In psychology, the ability of those who are administering tests or experiments to subtly, and even unknowingly, convey information about the answers to subjects is known in psychology as the “Clever Hans Effect.”

The question then arises as to whether investigators can give such subtle cues to witnesses, albeit unknowingly, and whether witnesses are as sensitive to them as Clever Hans. At least, the investigator should be very careful not to provide subtle cues. A deeper solution to this problem would be for lineups to be administered by investigators who do not know which of the exemplars is the suspect. A comprehensive analysis of the Clever Hans history, along with experiments that illustrate its applicability to human subjects is available in Rosenthal, Robert, 1998.

Similarly, investigators’ unintentional cues (e.g., body language, tone of voice) may negatively impact the reliability of eyewitness evidence. Psychology researchers have noted that such influences could be avoided if “blind” identification procedures were employed (i.e., procedures conducted by investigators who do not know the identity of the actual suspect). However, blind procedures, which are used in science to prevent inadvertent contamination of research results, may be impractical for some jurisdictions to implement. Blind procedures are not included in the Guide but are identified as a direction for future exploration and field testing. In the interim, an enhanced awareness on the part of investigators of the subtle impact they may have on witnesses will result in more professional identification procedures.
( return to top )

People v. Jason Richardson, 07ZF0016

Analysis and Critique of procedures

George Papcun
March 2010

According to Tustin Police Department Investigative Report CR#07-1052, p. 5, Detective M. Van Cleve #1054 on 03/23/2007, Detective Van Cleve states “I took an audio excerpt from Richardson’s interview at TPD on the date of his arrest as the base of the voice lineup. I used the statement, ‘She works for an environmental agency called Three E companies’ that Richardson made as the base. Detectives Mayfield, Sauerwein, and Alvarado along with Volunteer Ray Leger and IT Specialist Raul Delgadillo were audio recorded making the same statement.”
Issue: The statement of the suspect was collected from his natural speech, whereas the statements of the fillers were not. These differences produced changes in style and fluency that are audible in the recordings. (Reference: Guidelines Introduction Paragraph 4 and Guidelines: A. Composing Voice Lineups Discussion: Producing the utterance material)

Issue: The suspect speaks with a definite dialect that differs markedly from the speech of the others. Using the International Phonetic Alphabet, I have made a detailed phonetic transcription of the speech. These differences can best be illustrated by playing the recordings along with the transcription. (Reference: Guidelines Introduction Paragraph 3 and Guidelines A. Composing Voice Lineups Discussion: Producing the utterance material)

Issue: Given their occupations, it would appear that the educational levels of the fillers were considerably above that of the suspect. (Reference: Guidelines Introduction Paragraph 1 and Guidelines A. Composing Voice Lineups Discussion: Uniformity of accent, dialect, educational level, and speech characteristics)

Van Cleve’s report continues “The voice statements of each subject were then audio recorded onto a CD ROM for the voice line up.” and “I played the voice line up on a lap top computer with speakers for the witnesses that heard the suspect’s voice on the date of the incident.”

Issue: The recording and playback systems on laptop computers typically are limited in terms of their ability to faithfully represent the range of human voice frequencies without distortion. It would be of interest to determine the recording range of the recording and playback system to determine whether it played the frequencies of the human voice in an undistorted manner. (Reference: Guidelines Introduction Paragraph 5 and Guidelines A. Composing Voice Lineups Discussion: Recording and playing back the lineup)

In this connection, note that the transcription of the interview with witness Lucas Vocos renders the phrase used in the lineup as “She works for an environmental agency called 3D Companies.” (rather than “3E Companies”) This may be taken to indicate that the recording and/or playback is relatively poor. (Reference: Guidelines Introduction Paragraph 5 and Guidelines A. Composing Voice Lineups Discussion: Recording and playing back the lineup)

Van Cleve’s report continues: “ Isiah Atuatasi (Home Depot Employee/receiving): Atuatasi said he believed the suspect was the #4 or #6 voice, but he couldn’t be positive about it.” On the voice lineup identification sheet, Atuatasi is recorded as stating “Can’t be positive. Between #4 and #6.” This statement indicates that Atuatasi is using a comparative standard rather than an absolute standard for judging the voices.

Issue: The standard for an identification in a lineup is an absolute standard rather than a comparative standard. Thus, the issue in an identification procedure is not whether a given voice sounds more or less like the other voices in the lineup. Rather, it is whether a given voice is the voice that was heard previously. (Reference: Guidelines D. Recording Identification Results Discussion: Contradictory or otherwise anomalous responses)

Witness Dennis Malone. (Transcript DR NO.: 07-1052, p. 9, ll 13, 14.) Malone states “It’s either three or four because it was a deep voice.” The principal determinant of how deep a voice sounds is the rate of opening and closing of the vocal folds. Accordingly, I measured the average rate of vocal fold vibration for each of the subjects. The results are as follows:
As is apparent from these measurements, voices 3 and 4 are the deepest.
Voice 1: 115 times per second
Voice 2: 122 times per second
Voice 3: 96 times per second
Voice 4: 97 times per second
Voice 5: 103 times per second
Voice 6: 107 times per second
(Reference: Guidelines Introduction Paragraph 1 and Guidelines A. Composing Voice Lineups Discussion: Uniformity of accent, dialect, educational level, and speech characteristics)

(Transcript DR NO.: 07-1052, ll 6-11) Following Malone’s identification of the suspect, Van Cleve states “Okay. The next thing I’m going to do is I can tell you that’s the person we have in custody that we’ve arrested for the crime and we’re pretty comfortable that that’s the suspect in the shooting and so it’s good that you picked him out because we’re very confident we got the right person.” Providing that kind of information to the witness is improper because there is always the possibility that some other suspect will be developed. Providing such information irretrievably compromises the witness for future independent identifications. (Reference: Guidelines D. Recording Identification Results Discussion: Confirmation as to suspect choice) In this case in particular it is improper as well because a voice lineup is immediately forthcoming. Voices and visual features are correlated with each other on parameters such as height, age and other features. Therefore, providing the witness information about visual identification may influence the subsequent voice identification.

From transcript DR NO.: 07-1052, p. 10, l 1, Malone identifies voice three in the lineup (which is not the suspect) as the voice he heard previously. Van Cleve responds Okay. Well, like I said, you don’t have to pick anybody out.” (Transcript DR NO.: 07-1052, p. 10, ll 3,4) and continues in transcript DR NO.: 07-1052, p. 10, ll 4,5 “I’d rather you not pick one unless you’re 100 percent sure…” It is absolutely improper for Detective Van Cleve to express his preference about the how certain the witness should be. The investigator must simply ask the witness how sure he is and let it go at that.

Moreover, taking Van Cleve’s comments in context, they may be interpreted as trying to dissuade Malone from his decision. If this is the case, it calls into question Van Cleve’s objectivity.

Witness Francesco Cerda. On his voice identification record sheet, Cerda is recorded as stating “100% certain (of his identification of the voice of #6 in the lineup.) Not any of others” Because he mentions the others, his statement may be taken as indicating that Cerda was using a comparative rather than an absolute standard. Additionally, Van Cleve states in his investigative report “I asked him how certain he was and he said ‘Pretty certain. All the other one’s they don’t sound familiar at all.’ And ‘I’m pretty sure that’s the guy I heard.’ Thereafter, Van Cleve reports that Cerda’s identification is “100%” certain. Taken in context, this progression indicates what we might call “certainty creep.”

Witness Randy Carrillo.” MALE DETECTIVE: -- we have to try. With everybody I’ve showed this photo lineup to, so many people now, --
RANDY CARRILLO: Yeah.
MALE DETECTIVE: -- some have, actually, identified him.“
It is improper to state to a witness what other witnesses have said, either by way of identification or elimination of voices (or pictures). (Reference: Guidelines C. Conducting the Identification Procedure Discussion: Stating what other witnesses have said) and further, the detective states:
MALE DETECTIVE: We’re very confident we got the right guy. (Reference: Guidelines C. Conducting the Identification Procedure Discussion: Confirmation as to suspect choice)

Witness Richard Cocklin. MALE DETECTIVE: We’re pretty sure we have the right guy, yeah. (Reference: Guidelines C. Conducting the Identification Procedure Discussion: Confirmation as to suspect choice)

Witness Nicole Roman. “It’s been awhile since I heard the voice. I couldn’t really remember what it -- and I know that the guy didn’t have an accent.” In fact, Voice 6 has an accent distinct from the other speakers. (Reference: Guidelines D. Recording Identification Results Discussion: Descriptions of the voice)

Witness Jason Sanchez. MALE DETECTIVE: Well, that’s all I really had for you. I just had to show you the photo lineup, and just to let you know the investigation’s going really good. You guys know we have a guy in custody, and the police department is a hundred percent certain we’ve got the right guy. (Reference: Guidelines C. Conducting the Identification Procedure Discussion: Confirmation as to suspect choice)













( return to top )