top of page

Whodunit?: Bringing justice with forensic linguistics

Today I am interviewing Joseph Devney, a forensic linguist in California.

Hello Joe. First of all, please tell us what a forensic linguist does. Frankly, when I hear the word "forensic", I tend to associate it with autopsies, fingerprints, and dental records. How does linguistics fit in with this?

Your first instinct is correct. The “forensic” part refers to examining evidence in an investigation. And that is the narrow meaning of the term: linguistic analysis of evidence in court or in a police investigation.

But it has a broader meaning that is a more accurate description of the breadth of the field. It covers the intersection of language and law. In fact, some people use the term “language and law” instead because it reflects this perspective better.

This broader definition is preferred in the field, because it includes the things that are actually discussed by researchers and practitioners. Not just evidence in trials, but things like jury instructions, detecting plagiarism, the strategies used in police interviews, and interpreting statutory language. Any situation in which language studies and the law intersect.

The International Association of Forensic Linguists has members who know about these areas and more. (

Do you work with the written or the spoken word? Which is more challenging to analyze and decode?

I came to linguistics from a career as a writer. So I am more comfortable with the written word. But I can work from recordings and transcripts. With recordings, of course, some words or even full sentences can be unclear. Still, if at all possible, I would want access to the original audio recording, rather than trusting that someone else’s transcript is accurate enough.

Linguists pay closer attention to the details of language use than most people. To a linguist, a pause, or a filler word like “um,” can be significant if I am performing discourse analysis. We want to make sure that those things are captured in the transcript. And that utterances that are unclear are noted in the transcript that way as well.

There is a sister discipline, though, called forensic phonetics. The study of phonetics is part of linguistics, but these people specialize in analyzing voices and speech and audio recordings rather than the written word. They know a lot about how the sounds of speech are formed in the vocal tract.

The forensic phoneticians have their own professional organization, the International Association for Forensic Phonetics and Acoustics. (

What are the areas of law you deal with?

Any case in which language evidence is involved. Well, language evidence that I think I can analyze and come to useful conclusions. I recently worked on a criminal case, analyzing both a short witness statement and a very long police interview.

Linguists also get involved in trademark disputes, copyright cases, and interpretations of contract language—areas with clear connections between language and law.

Who employs forensic linguists?

In most cases, forensic linguists have “regular” jobs (often as linguistics professors), and do consulting in this field. The linguists can help with investigations—narrowing down a list of suspects for the police, for example. Tim Grant of Aston University in the U.K. (who just finished his term as president of IAFL) helps train police so they will be better interviewers. (

The FBI has had its own linguistic profilers on staff. The first of these, James Fitzgerald, is the one who used language evidence to get the search warrant that led to the arrest of Ted Kaczynski, the “Unabomber,” back in the 1990s. He has left the FBI, and is one of several forensic linguists who work independently.

But the typical client is a lawyer who needs help on a particular case. The linguist will come in to analyze whatever language evidence is in dispute, written or spoken. Is this trademark really a common word? Does this statement that seems to amount to a confession really mean that in the larger context? Are these passages from different books similar enough that they amount to copyright infringement? Could this defendant have written the ransom note that the police claim he did? Was what the defendant said on a recorded phone call a business proposition or a disguised threat?

That brings up another question. If a lawyer hires a forensic linguist, doesn’t that mean the linguist is there to build the case the lawyer wants, so they will win? Could the linguist bias his or her analysis that way?

I have heard this question before. The answer is no, not if the linguist is ethical. And we do have a code of ethics to follow, published by the Linguistic Society of America. A linguist submitting a report to the court can reference this code of ethics in the report. It says, for example, that the linguist won’t take cases in which the fee is dependent on the lawyer winning the case.

As an expert witness, the linguist’s obligation is to the court, not to the lawyer paying the fee. We analyze the language evidence, and simply report what we find. It may turn out that I need to tell my client, the lawyer, “this document supports the other side’s case.” The lawyer then makes her own decision about whether to use my report in court. To avoid a conflict of interest, I get paid the same fee no matter what I find out from my analysis.

What is your strategy to cope with people who not only speak a different language, but also come from cultures that are radically different from ours? I'm thinking of Mexican indigenous migrants from places like Oaxaca.

This would be a difficult problem. I assume that is why you are asking. Clearly, I would not have the requisite background to interpret testimony from people who don’t speak English, and few people would know those languages. I would start contacting people in the broader community of linguists, and try to find a linguist—or possibly a sociologist—who has worked with this language or this ethnic group.

You work in the U.S., and have mentioned the U.K. Is forensic linguistics practiced in other parts of the world?

Oh, yes. This summer I attended a conference sponsored by the IAFL, and met people from around the world. A woman from Kenya has been researching the metaphors that are used to just barely disguise the hate speech that helped incite election-time violence there. A linguist in Brazil teaches hostage negotiators. One speaker talked about her research into the way that juveniles are treated in the justice system in the Netherlands. I learned about how the use of English in the courtroom affects the delivery of justice in Hong Kong.

It really is used worldwide.

What is corpus linguistics and how can it be used as a tool?

A corpus in this context is a large sample of language from the “real world.” (The plural is corpora.) Corpus linguistics is the practice of analyzing these corpora using software. There are a lot of ways that corpora can be used in the legal context.

Recently there was an opinion piece in the Washington Post about using corpus linguistics to find the everyday meaning of words that might be relevant in a particular case. ( The authors write, “When problems of ambiguity in legal language arise, we expect judges and lawyers to turn increasingly to data from linguistic databases and less to mere dictionaries.”

An important use of forensic linguistics is identifying the author of an anonymous text—think of a ransom note or a terrorist threat. What can you find out about the writer from the way he or she uses language? Related to this task is author attribution, which means determining if a particular piece of unsigned writing could have been written by, for example, the defendant charged with writing it.

In both of these cases, corpus linguistics could be of use. For the anonymous threat, you might want to have a corpus of other “criminally oriented communications” (the FBI’s term) to sift through. The FBI has such a database; other consultants may have built their own. The linguist would do a search through the corpus to find similar documents, and see what clues that might provide.

The evidence in, for example, a corporate embezzlement case might include a couple hundred thousand e-mail messages. You would need sophisticated search software, and the skills of a linguist, to analyze that corpus and identify the minority of messages that relate to the case, and start constructing timelines and networks of people.

What is the role of translating and interpreting in your work? What are the consequences of a poor translation, written or spoken?

I work in English, though I do have some knowledge of Spanish. If having a foreign language in the mix is relevant to the analysis needed, I would contact people with the right backgrounds. I have a colleague in Quebec who has already told me to call her if a case involves language evidence in French. I worked on a case in which a bit of the Spanish language was relevant. I did my own analysis, and then reached out to the forensic linguistic community. I discussed the case, and the Spanish quote in particular, with a linguist who is also a court interpreter. She gave me useful information, and I quoted her in my report.

A lot of forensic linguists are concerned in particular with issues related to translation and interpretation, and to the problems that non-native speakers face in the justice system. Here is a good example. One speaker at the IAFL conference, Michael O’Laughlin of Boston University, is a court interpreter and expert witness for language issues. In his conference presentation he showed a video of a murder suspect whom he identified only as “Pedro.” Let me tell the story as Michael described it to me.

Pedro was given a Miranda Warning both in English and in Spanish, but the policemen used as interpreter spoke limited Spanish and evidently could not read Spanish very well. This officer was a “heritage speaker,” meaning that he was the son of immigrants who spoke Spanish, but his main language was English. Although he garbled the Miranda Warning he read in Spanish, mispronouncing many of the words, Pedro said “yes” every time he was asked if he understood.

Dr. O’Laughlin, in his capacity as expert, needed to show that Pedro did not understand the Miranda Warning, even if he said yes over and over. He did this by also presenting Pedro’s Waiver of Prompt Arraignment, which was also read out by the Spanish-speaking officer. This form had been very badly mistranslated, and could hardly be understood. It also contained words, such as “arraignment,” which were left in English. The Spanish-speaking officer should have told his superiors that the form made no sense. Instead, he tried to read it, and added another layer of incomprehensibility due to his mispronunciations. The funniest mistake was the word for “release.” Instead of “liberado,” the form read “lanzado,” which means to shoot into the air. Because the Spanish-speaking officer could not pronounce this unknown word, he said “losado,” which, if it means anything, means covered in floor tiles! Why was this form important? Because Pedro said yes when asked if he understood this form, when in fact no one in the world could have understood it. It proved that Pedro answered yes even when he had little clue of what was being said, and that was very important for critiquing his understanding of the Miranda Warning.

In a case that another speaker told about, a non-native speaker was not provided with a translator. There was no formal evaluation of his English skills. The police had a casual conversation with him, and decided, “Oh, he doesn’t need a translator.” Being able to contribute to a simple conversation is not at all the same as following what is going on in a police interview or a court proceeding. As the Miranda warning says, “anything you say can be used against you.” And possibly taken out of context.

How do you envision the future of forensic linguistics? Will it play a larger role?

I do hope it plays a larger role. Linguists, who study in detail how language works, have a lot to contribute to the legal system, in the U.S. and in other countries. Part of my role, as I see it, is to educate people about how linguists can help to bring more justice into the world.

Beyond that concern, one trend I see in the field is more reliance on software and databases. As I mentioned earlier, corpus linguistics can be of help in investigations. And at the IAFL conference I mentioned, Malcolm Coulthard of Aston University in the U.K. talked about the future of the field, and he sees something similar. (Here I will refer to my own notes, and offer my apologies to Dr. Coulthard if I have mischaracterized what he said.) He thinks it is likely that there will be a “Google Translate-type” software for legal translation and interpretation. He expects computerized authorship attribution—something I alluded to earlier, when I mentioned the databases and software that some consultants have been building. He also envisions new ways of analyzing language to determine if someone is trying to assume someone else’s identity.

There seems to currently be an “arms race” between people developing software to obscure plagiarism, and software to detect it, based on what I heard from another researcher who spoke at the IAFL conference. I don’t know who will win.

And there is another trend I see, very different. I know some forensic linguists who are devoting more energy to the problem of false confessions. Surprisingly, people confess to crimes that they did not commit. Sometimes it is because they feel trapped, with no way out except to confess. Or it may be that they have been convinced that they did commit them during police interrogations. It’s tragic. Psychologists have known for some time that it is not terribly hard to implant a false memory in someone. Linguists are working to figure out when confessions are not authentic, and what to do about it. This is one of many ways that forensic linguists are trying to bring more justice into the world.

Thank you for your time, Joe. It's been an education!

Featured Posts
Recent Posts
Search By Tags
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square
bottom of page