Forensic linguistics is the application of linguistic knowledge and methods to legal questions. It covers everything from identifying who wrote an anonymous threatening letter to analyzing whether a contract’s wording misled consumers. The field sits at the intersection of language science and the justice system, and practitioners work on both criminal and civil cases.
What Forensic Linguists Actually Do
The scope of the field is broader than most people expect. The International Association for Forensic and Legal Linguistics recognizes several major areas of work: analyzing the written language of laws and legislation, studying spoken legal discourse in courtrooms and police interrogations, providing linguistic evidence on identity or authorship, addressing social justice issues that emerge from legal language, and consulting on legal drafting and interpretation. A forensic linguist might spend one week comparing two documents to determine if the same person wrote them, and the next week testifying about whether a police officer’s phrasing during an interrogation was coercive.
Some practitioners focus entirely on written text. Others specialize in speech and audio recordings. A smaller but growing number work on digital communications: texts, emails, and social media posts. What unites them is the core principle that language contains patterns, and those patterns can be measured, compared, and presented as evidence.
Authorship Attribution
One of the field’s signature tasks is figuring out who wrote a given text. The underlying idea is that every person develops individual language habits over time, many of them unconscious. These habits form a kind of linguistic fingerprint. Some features are so distinctive that the probability of another person reproducing them is considered negligible.
Linguists examine several layers of a text to build an authorship profile. At the vocabulary level, they look at word length, sentence length, vocabulary richness, the frequency of function words (like “the,” “of,” “but”), and even the relative frequency of long words (those with more than six characters). Function words are particularly useful because writers choose them without thinking. You might consciously decide to use the word “surrogate” in an essay, but you don’t consciously decide how often you use “however” versus “nevertheless.”
At the grammatical level, analysts break text into parts of speech and look for recurring syntactic patterns. The idea is that every author unconsciously favors certain sentence structures. Punctuation habits, paragraph organization, and spelling tendencies all contribute to the profile. When analysts have a large enough sample, these features can reliably narrow down or confirm authorship.
The Unabomber Case
The most famous example of forensic linguistics in action is the identification of Ted Kaczynski as the Unabomber. FBI profilers initially believed the bomber was young and poorly educated. Linguistic analysis told a different story.
James Fitzgerald, the FBI’s forensic linguist on the case, noticed that Kaczynski’s manifesto used words like “surrogate,” “overspecialization,” and “tautology,” pointing to a high level of education. His word choices also revealed his age. He referred to women as “broads” and “chicks” and used the word “negro,” terms that had fallen out of common use by the early 1990s when the manifesto was published. Phrases like “holy robots,” “working stiff,” and “playing footsy” were characteristic of someone who grew up in the 1960s.
Linguist Roger Shuy identified dialect clues as well. Kaczynski’s writing contained terms associated with Chicago-area newspapers from the 1930s through the 1950s. He also spelled “willfully” as “wilfully” and “clue” as “clew,” misspellings that matched spelling reforms adopted by The Chicago Tribune during the 1940s and 1950s. In letters accompanying some of his bombs, Kaczynski mentioned going out “in the sierras” to relax. Using “sierra” as a common noun rather than a proper noun for the Sierra Nevada range is a habit specific to people from the western United States, particularly Northern California. Even his punctuation, including the use of two spaces between sentences, reflected an older writing style consistent with his age and education. Taken together, these details helped build the case that led to his arrest.
Speaker Identification From Audio
Forensic linguistics isn’t limited to written text. A major branch of the field deals with identifying speakers from audio recordings, often in cases involving threatening phone calls, ransom demands, or disputed recordings.
Several methods are currently in practice. Auditory phonetic analysis involves an expert listening carefully to speech samples and attending to features at both the individual sound level and broader patterns like rhythm, intonation, and stress. Acoustic phonetic analysis uses computer software to measure the physical properties of the speech signal, quantifying things like pitch patterns and the resonance characteristics of a speaker’s voice. Most practitioners combine both approaches, using their ears alongside digital tools.
More recently, automatic and semi-automatic speaker recognition systems have entered the field. These use algorithms to compare voice samples, sometimes with human oversight. The most thorough approach combines human auditory and acoustic analysis with automated systems. An older technique called “voiceprint” analysis, based on visual patterns from a device called a spectrograph, has largely fallen out of favor. Its main flaw was that it reduced voice comparison to simple pattern matching without giving examiners access to the actual audio, obscuring the phonetic nature of what they were analyzing.
Analyzing Threats and Suicide Notes
Forensic linguists also assess the language of threat communications and suicide notes. In threat cases, the goal is often to determine whether a threat is credible and to profile the person who made it. In death investigations, analysts may be asked to determine whether a suicide note is genuine or forged.
Research on suicide notes has identified consistent patterns in their content. In one study of over 100 notes, roughly 39% were primarily farewell messages. About 27% focused on explaining the reason for the suicide. Around 16% left instructions about practical matters like the distribution of belongings, and another 16% asked for forgiveness or apologized. Notes expressing affection appeared in about 8% of cases, while those assigning blame to others appeared in about 5%. The presence of a note generally indicates that the person remained psychologically connected to others at the time of writing, using the note as a final opportunity to explain, apologize, or offer comfort. Gender differences have also been observed: in one study, no women’s notes contained apologies for the act, and only one expressed explicit affection, compared to higher rates among men. These patterns give forensic linguists a baseline for evaluating whether a note’s language and content are consistent with a genuine suicide.
Legal Language and Contract Disputes
Not all forensic linguistics involves crime. Linguists regularly consult on civil cases where the meaning of a law, contract, or warning label is disputed. This work involves analyzing ambiguity, readability, and whether ordinary consumers could reasonably understand the language they were presented with. A forensic linguist might be asked whether a product disclaimer was written clearly enough to constitute informed consent, or whether a clause in an insurance policy has two plausible interpretations. They also assist in plain language drafting, helping legal professionals write documents that are less likely to be misunderstood or challenged in court.
Detecting AI-Generated Text
A newer challenge for the field is distinguishing between text written by a person and text generated by artificial intelligence. In a controlled study where experts were asked to identify AI-generated medical essays mixed in with human-written ones, participants correctly identified the AI text 70% of the time. Medical experts and humanities experts performed similarly (72% versus 65%), suggesting that familiarity with the subject matter didn’t provide much advantage.
The most useful clues were stylistic rather than factual. Redundancy made experts nearly seven times more likely to flag a text as AI-generated. Repetition had an even stronger effect, making identification about eight times more likely. Issues with coherence, where the logical thread of an argument felt off, also played a significant role. Content errors, surprisingly, had little impact on whether experts identified a text as machine-written. The decision-making process relied primarily on linguistic attributes rather than factual accuracy.
How Courts Evaluate Linguistic Evidence
In the United States, expert testimony from forensic linguists is evaluated under a legal framework established by the Supreme Court. Judges assess whether the expert’s methodology can be tested, whether it has been published and peer-reviewed, what its known error rate is, whether standards exist for its application, and whether it has gained widespread acceptance within the relevant scientific community. This framework, originally designed for scientific testimony, was later extended to cover non-scientific expert testimony as well. Opposing attorneys can challenge a forensic linguist’s testimony before trial through pretrial motions, arguing that the methods used don’t meet these reliability standards.
This means forensic linguists need to demonstrate that their analytical methods are systematic and reproducible, not just intuitive judgments about how a text “sounds.” The field has responded by developing more rigorous quantitative methods alongside traditional qualitative analysis.
Training and Professional Development
Forensic linguistics typically requires graduate-level education. Hofstra University, one of the few institutions offering a dedicated master’s degree in the field, requires 36 semester hours: 21 hours of core coursework, 12 hours of electives, and a master’s thesis. Full-time students can finish in two years, while part-time students generally take three. Many practitioners come from broader linguistics or phonetics backgrounds and specialize through additional training and casework experience.
The primary professional organization is the International Association for Forensic and Legal Linguistics, which hosts international conferences and publishes the International Journal of Speech, Language and the Law. Membership starts at $80 per year, with discounted rates for students and scholars from the Global South. The field remains relatively small compared to other forensic disciplines, but demand for qualified practitioners has grown steadily as digital communication generates more text-based evidence than ever before.