Why 911 Call Analysis Keeps Sneaking Into Your Cases

You probably didn't see it labeled in the discovery index. No police report mentions "call analysis." No witness list names a forensic linguist. But somewhere in the stack, in a transcript of a detective's interview, in an officer's supplemental report, in the CV of a witness the State just endorsed, there's a fingerprint of 911 call analysis. And it's reaching your juries.

A new Cornell Law Review article calls this pattern "shadow forensics": forensic conclusions that reach the jury without ever crossing the Daubert or Frye threshold. The trick is simple. The underlying methods are never disclosed. The founders never take the stand. Instead, a 911 operator or a detective who attended an eight-hour, $3,500 training session testifies as a lay witness about whether the caller sounded guilty.

ProPublica documented more than 100 cases across 26 states where this happened. Of roughly 50 agencies and training associations ProPublica canvassed, only 11 responded, and most of those distanced themselves from the program. Some agencies didn't even know their own personnel had been trained.

The pipeline is the part defenders miss. Departments pay up to $3,500, often out of taxpayer funds, for an eight-hour course that promises to identify the guilty caller through word choice, phrasing, cadence, and grammar. The original study behind the method examined only 100 calls and was meant as exploratory research. At least eight replication studies have failed to validate it. None of that stops the trainings, and none of it stops the testimony.

The mechanism matters because it changes what you challenge. You aren't fighting an expert opinion under Rule 702. You're fighting a lay witness who happens to be repeating one. The objection lives in the substitution, not the science.

This is the discovery problem you didn't know you had. When you pull the 911 audio, the call-taker notes, and the officer's supplemental, you're already sitting on the raw materials for a 911 call transcript analysis that can surface the operator's conclusions before they reach opening statements. Pair it with jail call transcription when the same witness later testifies about a defendant's post-arrest demeanor, and you've got the report-versus-audio inconsistency that breaks the substitution on cross.

What the Science Actually Says (and Why It Fails)

The science behind 911 call analysis doesn't hold up. It never really did.

The technique rests on an original study that examined only 100 calls and was intended as exploratory research, not as a foundation for real-world criminal prosecution. One of the study's own co-authors later told journalist Brett Murphy the findings were not ready for field use. Since then, at least eight replication studies have failed to validate the guilt indicators the method claims to detect. The 911 COPS Scale, the main instrument used to assess caller credibility, lacks "robust empirical support" for the conclusions drawn from it.

The deeper problem is what Dr. Jessica Salerno calls the emotional-display fallacy. Her research shows that what investigators read as "guilt" is often just the absence of expected emotion. Officers in the studies required callers to sound highly emotional before suspicion decreased; moderate distress was not enough. Trauma, culture, personality, and neurodivergence all shape how a person responds during a crisis, yet the criminal legal system keeps treating grief and shock as if they come in one recognizable package.

That gap between expectation and reality has destroyed lives. Huwe Burton was 16 when he called 911 to report his mother's murder. An officer concluded he "did not seem emotional enough." Burton served more than 20 years before his exoneration in 2019. Robert Roberson spent more than 20 years on death row after his flat affect during a 911 call was read as evidence of guilt; he was later diagnosed with autism spectrum disorder, and he narrowly avoided execution three times.

These aren't edge cases. They show exactly how the method's assumptions fail in court. The harm shows up in concrete ways:

• The original study examined only 100 calls and was never meant for field use.
• At least eight replications have failed to validate the technique's guilt indicators.
• The 911 COPS Scale lacks robust empirical support for credibility findings.
• Officers in Salerno's research required high emotional display before suspicion dropped.

For defenders, the takeaway is practical: when a detective or dispatcher testifies that a caller "didn't sound right," that conclusion is built on a 100-call exploratory study, a failed-to-replicate instrument, and a debunked assumption about how real people react in crisis. The audio itself, properly transcribed and timestamped, lets you put the call in front of a jury and let them hear what the officer claims to have heard. If you're building a case file around a 911 call, 911 call transcript analysis gives you a diarized record with word-level timestamps so you can challenge each characterization at the exact moment it was made.

How to Challenge the Lay-Witness Introduction in Your Case

The "shadow forensics" mechanism described in the Cornell Law Review explains why this technique often slips past standard evidentiary hurdles. Prosecutors avoid expert testimony by having detectives or dispatchers offer opinions as lay witnesses. This substitution pattern allows subjective conclusions to enter the record without scientific validation.

Your first task is spotting this tactic in discovery. Look for police reports that reference "call analysis" training or consultations with specific analysts. Check if the testifying officer attended a workshop rather than holding a scientific degree. The underlying method is never disclosed, and the founders are never proffered as experts, which is the core of the problem.

Dr. Jessica Salerno’s research highlights the danger of this approach. Investigators often interpret a lack of high emotional display as guilt. In one case, a sixteen-year-old was convicted because an officer felt he "did not seem emotional enough" during a crisis call. Another defendant spent over twenty years on death row because his flat affect was misread as deception, a trait later linked to autism spectrum disorder.

To build a motion in limine or Daubert challenge, you need to isolate the undisclosed method. If the officer mentions specific "indicators" from a training they cannot explain, you have your hook. You are challenging the reliability of the process, not just the opinion.

Use this checklist to frame your objection:

• Identify if the witness claims expertise based solely on a short training session.
• Note any references to proprietary scales or specific "guilt indicators" without scientific backing.
• Cross-reference the officer’s testimony against the raw audio to find subjective interpretations.
• Cite the lack of peer-reviewed validation for the specific method being used.

The 911 COPS Scale© lacks robust empirical support, yet it is routinely used to assess credibility. At least eight replication studies have failed to validate the method's supposed guilt indicators. When you present this data, you shift the burden to the prosecution to prove the technique's reliability.

Reviewing the raw audio is essential for this process. You need to hear exactly what the caller said and compare it to the officer’s written interpretation. A specialized service for 911 call transcript analysis can provide timestamped, diarized transcripts that make these discrepancies obvious. Similarly, reviewing jail call transcription helps track how narratives evolve from the initial call to later interrogations.

Evidence Copilot supports this workflow by allowing you to upload the 911 audio and police reports side-by-side. The platform surfaces inconsistencies between the report and the call audio as review signals. This grounded approach ensures every point in your motion is tied to a specific, citable moment in the evidence.

You can book a 20-minute walkthrough at evidencecopilot.com/book to see how this workflow fits into your case preparation. The tool is designed to help you spot these issues faster, but the legal strategy remains yours.

Post-Conviction Review and Institutional Responses to Track

If your client is already convicted, the 911 call analysis technique may have influenced the verdict without you realizing it. The stakes are high because faulty scientific testimony is the second most common factor in wrongful convictions. You need to know how to audit closed cases for this hidden evidence.

ProPublica documented over 100 cases across 26 states where law enforcement used the technique. This volume suggests that many closed files contain the same evidentiary flaws. The Illinois Supreme Court recently agreed to review Jessica Logan's conviction, signaling that appellate courts are taking these claims seriously.

Institutional responses are shifting from denial to accountability. North Carolina’s Office of Indigent Defense Services issued a statewide warning to attorneys. This move helps defenders identify potential issues in their own jurisdiction. Fair and Just Prosecution also called on member prosecutors to review past cases. They argue that prosecutors must correct past injustices through post-conviction review processes.

To determine if the technique touched your client's closed case, look for specific markers in the discovery. Check for references to call analysis training or consultations with the method’s creators. Review officer testimony that discusses caller culpability based on phrasing or cadence. If you find these elements, you have a basis for a motion to review.

The technique often enters through lay witness testimony rather than expert testimony. This "shadow forensics" approach allows it to escape standard Daubert challenges. Officers testify about what they heard, bypassing the need for scientific validation. You must spot this substitution pattern to build a successful post-conviction argument.

Reviewing the underlying audio is critical for this process. You need to compare the officer's narrative against the actual recording. Tools that provide timestamped diarized transcripts help you find discrepancies quickly. For instance, 911 call transcript analysis allows you to pinpoint exact moments where the officer’s interpretation diverges from the caller’s words. This precision strengthens your motion for review.

Similarly, jail calls often contain statements that contradict earlier police reports. Jail call transcription helps you surface these inconsistencies before they get buried in the file. Every piece of evidence should be grounded in citable sources. This approach ensures your arguments are fact-based and ready for judicial scrutiny.

• Search discovery for mentions of "call analysis" or "Harpster" consultations
• Identify officer testimony that infers guilt from caller behavior
• Compare officer narratives against the raw 911 audio timestamps
• Document any discrepancies between the report and the call content

The Cornell Law Review has provided the legal framework for attacking this technique. It explains how the method evades evidentiary oversight. Use this scholarship to support your post-conviction motions. The goal is to ensure that no innocent person serves time based on junk science.

If you need help reviewing complex audio files, consider booking a walkthrough. Evidence Copilot offers a grounded evidence-review workspace for defense teams. You can start with a 20-minute walkthrough to see how the platform works. Or, you can start on the Solo plan at evidencecopilot.com/signup. Every AI answer cites the exact source moment, helping you build airtight arguments.

Anticipating the Rebuttal and Reviewing the 911 Audio Yourself

Tracy Harpster doesn't concede the premise. He denies that 911 call analysis produced any wrongful convictions, says he examined more than 2,500 equivocal death cases over roughly two decades, and argues the technique has been used to exonerate people, according to reporting on the prosecutorial pushback. He and Susan Adams have also argued that ProPublica's coverage presented an inaccurate narrative and omitted key evidence in the Logan and Helmick cases.

Know that rebuttal before you walk into a hearing, because you will hear it. The stronger position isn't that Harpster is lying. It's that the record is contested and the method was never subjected to the scrutiny that would settle the question. Cornell Law Review's "shadow forensics" framing gives you that ground: the technique escapes Daubert and Frye because its methods are never disclosed and its founders are never proffered as experts. Instead, operators and officers testify about caller culpability after attending a training.

That's your objection to build, and it starts in the discovery itself.

• Pull the 911 audio and the police report and read them side by side, not sequentially. The report narrates the call; the audio is the call.
• Flag every place the report characterizes the caller's tone, emotion, or demeanor. Those adjectives are where the analysis enters.
• Check whether any officer or dispatcher on your witness list has call-analysis training. Ask for the training materials and the underlying study.
• Note the gap between what the caller actually said and what the report says the caller's words meant.

The practical problem is scale. A single case can carry hours of 911 audio plus jail calls, and the inconsistencies you need live at the word level: a caller who says "I found him" versus a report that reads "I discovered the body." 911 call transcript analysis with word-level timestamps and speaker diarization lets you search the audio instead of replaying it, and jail call transcription does the same for custodial calls. When you upload the report and the audio to the same case, the workspace compares new evidence against what's already there and surfaces report-versus-call inconsistencies as review signals before you ask. Every AI answer cites the exact moment it came from, and every output is marked for human review, because the judgment about what that moment means stays yours.