Do You Hear What I Hear?

Part II
A Research Study into the Interpretation of EVP

A Second Study of EVP Interpretation
by Mark Leary, Ph.D.

Published in the Spring 2013 ATransC NewsJournal
Read Part 1 and Part 3

This is a Minibox by Paranormal Systems, Inc. (2007). It is basically a radio equipped with a tuner that automatically sweeps the dial. The sweep is continuous and the rate is variable. See the ATransC endnote.


In the previous issue of the NewsJournal (Winter 2013), I described a research study that examined the problem of EVP interpretation. As all EVP enthusiasts know, people often disagree over how particular EVP should be interpreted, and we were interested in documenting how serious the problem really is.

In that study, 24 investigators each interpreted a large set of EVP, and the most common or “consensus” interpretation of each EVP was determined. Then, the individual investigators’ interpretations of the EVP were compared to the consensus interpretations to see how well they agreed. The results showed that, on average, only 21% of the investigators’ interpretations of particular words agreed with the consensus interpretation. To put this finding in perspective, imagine that your family doctor arrived at the same diagnosis as most other doctors on only 1 out of 5 of their diagnoses. Such a low rate of agreement would obviously raise serious issues about medical diagnoses, and similar issues must be addressed about EVP interpretations.

In this article, I describe a second study that was conducted to answer additional questions about EVP interpretation. This study, which was partly funded by a grant from the Association TransCommunication to the Rhine Research Center, examined EVP that were recorded using radio-sweep technology. Radio-sweep technology, often known as “ghost boxes” or “spirit boxes,” involves rapidly changing the tuning of a radio receiver to produce a stream of noise that is composed of bits of sound from the stations that are being scanned. Advocates of this technique believe that communicating entities use the snippets of sound to produce words. Many investigators suggest that EVP that are recorded with radio sweep are more distinct than those recorded without background noise. However, critics note that the noise source itself sometimes contains words or other sounds that might be interpreted as intelligent communication. In any case, we were interested in whether the low rate of agreement found in the earlier study of EVP that were recorded without a sound source is also found with radio-sweep EVP.

A second goal of the study was to examine how people’s interpretations of EVP are affected by knowing what other people heard. EVP enthusiasts know that people’s interpretations of EVP can be influenced by what other people say they hear. For that reason, some investigators do not share their personal interpretations until others have listened and come to their own, independent conclusions. But exactly how much are listeners’ interpretations biased when they know what other people think an EVP says? And does this biasing effect depend on whose interpretation is known? Often, listeners tend to give the interpretation offered by the investigator who recorded the EVP special attention, possibly because listeners assume that the original investigator has listened carefully many times before rendering an interpretation and is aware of the conditions under which it was recorded. If so, listeners may be particularly affected by knowing the interpretation of the person who recorded the EVP. To examine the biasing effects of knowing other people’s interpretations, we had people listen to EVP after learning what others thought they said and had other people interpret EVP without knowing others’ interpretations.

The Study

To obtain a set of EVP for the study, an announcement was posted on the ATransC website and published in the NewsJournal asking investigators to submit radio-sweep EVP that the investigator believed contained an anomalous or ethereal voice. We also asked the submitting investigators to indicate what they thought the EVP said.

Nineteen EVP were submitted, of which we selected 12 audio clips for the study. If an investigator’s statement or question preceded the clip, it was removed so that each clip contained only the EVP with a few seconds of radio-sweep noise before and after when possible. These 12 EVP varied in length from a one-syllable word to eight words (containing 11 syllables).

Before we started the study, a pair of experienced paranormal investigators provided their independent interpretations of each EVP. We used these interpretations to see whether people’s interpretations agreed more with the original investigators’ interpretations than the “secondary interpretations” provided by these other investigators

The original investigators’ interpretations of the 12 EVP contained a total of 55 syllables and 46 words. The secondary interpretations by the other investigators contained 63 syllables and 53 words. The original and secondary interpretations not only contained different numbers of syllables and words, but the secondary interpretations agreed with the investigators’ interpretations on only 4 words (8.7%).

Ninety adults were recruited to participate in the study. The participants were 23 men and 61 women who ranged in age from 18 to 81 (average age was 46.5). Dr. Christine Simmonds-Moore supervised the data collection, which occurred in a laboratory at either the Rhine Research Center or the University of West Georgia.

After each participant completed a background questionnaire, he or she was randomly assigned to one of three experimental conditions. These experimental conditions differed in whether participants received an interpretation of each EVP before listening to it. Participants in the no-interpretation condition simply wrote down their interpretations on a form that we provided, one EVP per page. Participants in a second condition saw the investigator’s interpretation of the EVP at the top of each page of the form before interpreting the EVP. Participants in a third condition saw the secondary interpretation (made by the other two investigators) at the top of each page. By having participants listen to the EVP under three different conditions (no interpretation, investigator’s interpretation, or secondary interpretation), we could examine the degree to which knowing others’ interpretations affected what participants reported they heard.

Participants listened to each of the 12 EVP clips through headphones in a quiet research room and wrote down their interpretations. Participants listened to each EVP as many times as needed to decide what the words might be.

Agreement with the Investigator’s and Secondary Interpretations

In deciding whether particular syllables and words in the participant’s interpretation matched the syllables and words in the investigator’s and secondary interpretation, we leaned in the direction of leniency. For example, singular and plural forms of a word were counted as a match (Richard/Richards), contractions were counted as a match with their constituent words (“don’t” and “do not” were counted as a match), and homonyms were counted as a match (weight/wait, their/there, hire/higher). Also, a particular word did not have to appear in the same position in the participant’s interpretation as in the investigator’s or secondary interpretation. For example, “book” would be counted as a match in “now take the book away” and “he should write the book today” even though book is the fourth syllable of the first phrase and the fifth syllable of the second.

The primary question was how many of the words in the investigators’ and secondary interpretations participants heard. Did participants hear the same things as the investigators? The answer depends on whether participants saw an interpretation of an EVP before they interpreted it.

When participants did not see any interpretation before listening to the EVP, they agreed with the investigator’s interpretation on only about 6% of the words (and 10% of the syllables) and with the secondary interpretations on about 8% of the words (10% of the syllables). More discouragingly, among participants who did not learn any interpretations before hearing the EVP, only one participant interpreted an EVP precisely in the same way as the investigator who submitted it. That is, out of 360 interpretations in the no-interpretation condition of the study (30 participants × 12 EVP), only one perfectly matched what the investigator reported. This is obviously a rather low level of agreement.

Of course, agreement differed across the 12 EVP. On the EVP with the greatest agreement, only one participant failed to match at least one word in the investigator’s interpretation. But on the EVP with the least agreement, not a single participant agreed with the investigator’s interpretation!

Agreement also differed across participants. In the absence of knowing the interpretation, the “worst” participant agreed with only 2% of the investigators’ words, and the “best” participant agreed with 14% of the words.

Knowing the investigators’ or secondary interpretations improved agreement markedly. Participants who had seen the investigators’ interpretations reported an average of 23% of the words and 23% of the syllables in those interpretations. After seeing the secondary interpretations, participants’ interpretations matched 27% of the words and 25% of the syllables in those interpretations. Thus, knowing how other people interpreted an EVP strongly influenced what participants said they heard.


Although the results of this study say nothing whatsoever about the nature of EVP, they raise questions about the degree to which we can trust interpretations of most EVP. I imagine that some investigators will find these results exceptionally discouraging. Most EVP enthusiasts are quite aware that people’s interpretations of particular EVP often disagree, sometimes wildly, but the extent of the problem may be more sobering than many imagine.

“Methods, the psychophone and the EVPmaker software methods proved to be highly unreliable, not because they are particularly bad acoustic backgrounds for the production of the voices but because they are undoubtedly a source of uncertainty and ambiguity in the analysis of the results. They can very easily originate pareidolia and/or projection of meaning based upon expectation. Very particularly with the EVPmaker software, it is easy to find “results” in recording-sessions where they do not exist. In addition, an erroneous interpretation of the content of possibly anomalous utterances found in the recording is very likely. Most of the EVP “results,” published in the Internet, fall into one of these categories.”

From: A Two-Year Investigation of the Allegedly Anomalous Electronic Voices or EVP

by Anabela CardosoRepublished from NeuroQuantology | September 2012 | Volume 10 | Issue 3 | Page 492-514

Other investigators may object that the situation is not as dire as the data suggest. For example, some may object that most of the participants in this study were not experienced with recording or interpreting EVP and, thus, the data may say little about the quality of EVP interpretations by experienced investigators. Yet, as noted, the two experienced investigators who provided the secondary interpretations agreed with the investigators’ interpretations on only 11.5% of the words. And the fact that participants who were trying to identify the words in the recordings under controlled circumstances heard only about 6% of the same words as the investigators should give us pause.

In many instances, these disagreements have no practical implications. If investigators on a paranormal investigation of a public location don’t agree on their interpretation of a particular EVP, often no harm is done. However, in cases where other people have a stake in the interpretation—as when grieving parents believe that an EVP is the voice of a deceased child—then interpretations matter a great deal. Like the earlier study, this experiment suggests that investigators should be more cautious in interpreting EVP for other people when the interpretations matter.

Read Part 1 and Part 3

learythumbnailsmallWith a Ph. D. in social psychology, Dr. Leary is a research psychologist who studies topics related to self-awareness, motivation, and emotion. He has conducted research on topics such as reactions to social rejection, the effects of excessive self-attention, people’s concerns with their social images, and the relationship between personality and behavior. He is on the editorial boards of several scientific journals in social psychology and recently released a psychology course on DVD entitled “Understanding the Mysteries of Human Behavior.”

Editor’s Note

For an additional study of how people hear EVP, please refer to the article EVP Online Listening Trials in the ATransC online Journal.

“Radio-sweep” is a generic name for EVP thought to be formed using sound produced by sweeping a radio dial. In principle, it produces a form of EVP referred to as “opportunistic EVP.” Please review Locating EVP Formation and Detecting False Positives and Radio-Sweep: A Case Study. Also see the article on page 9: “A Two-Year Investigation of the Allegedly Anomalous Electronic Voices or EVP.”

Leave a Reply