History and Landmark Cases: Svartvik, Evans, and Coulthard

Timothy John Evans lived at 10 Rillington Place in Notting Hill, London, with his wife Beryl and their infant daughter Geraldine. In November 1949, both were found strangled. Evans was arrested, confessed, and in March 1950 was hanged for Geraldine's murder. The confession was the prosecution's central evidence.

In 1953, the bodies of six women were found at the same address, concealed in the walls and under the floorboards. The tenant, John Reginald Halliday Christie, was identified as a serial killer. The Brabin Report (1966) concluded it was more probable than not that Christie had murdered Geraldine Evans, not Beryl, and that Evans himself had more probably than not killed Beryl. Evans's case was reviewed repeatedly. He received a posthumous free pardon in 1966 under pressure from campaigners including journalist Ludovic Kennedy, whose book 10 Rillington Place (1961) documented the evidence of wrongful conviction.

In 1968, Svartvik published his analysis of the Evans statements. He divided the statements into sections linguistically and showed that those parts attributing the murders to Evans contained a range of features, vocabulary choices, syntactic constructions, discourse markers, that did not appear in the portions of the text attributed to Evans's own narration. The contrasting style was consistent with police officers having inserted or rewritten the incriminating passages. Svartvik was careful: he did not claim the statements were definitely fabricated. He argued that the textual evidence was consistent with that interpretation and inconsistent with the statements being a uniform record of Evans's own speech.

Malcolm Coulthard, a linguist at the University of Birmingham, brought a sharply corpus-based approach to several high-profile British miscarriage-of-justice cases in the 1980s and 1990s. His work on the Birmingham Six, six Irishmen wrongly convicted of the 1974 IRA pub bombings in Birmingham, and on Judith Ward, wrongly convicted of the 1974 M62 coach bombing, centred on the same question Svartvik had raised: were the confession statements in the defendants' own words?

Coulthard's method was to look not just at vocabulary but at function words and discourse structure, the small words and organisational patterns that speakers use without thinking and that are therefore harder to fake or suppress. He noted in the Birmingham case that police officers routinely used a particular word in a way that native English speakers almost never do in spontaneous speech: the word 'then' as a discourse marker linking clauses. The pattern appeared in the attributed statements at frequencies consistent with police drafting practice, not with informal speech.

Coulthard was also prominent in the case of Derek Bentley, hanged in 1953. The disputed phrase 'Let him have it, Chris', attributed to Bentley and treated by the prosecution as an instruction to shoot, became the subject of pragmatic analysis. Does the phrase mean 'hand the gun to the officer' or 'shoot him'? Coulthard's argument that the phrase was genuinely ambiguous and that its incriminating reading required assumptions no fair jury should be required to make contributed to Bentley's posthumous royal pardon in 1993, with his murder conviction formally quashed by the Court of Appeal in 1998.

Roger Shuy was a sociolinguist at Georgetown University who began taking on casework from the 1970s and published widely on the practical results. His cases were mostly American criminal prosecutions: bribery and corruption cases relying on undercover recordings, conspiracy charges, extortion allegations. The linguistic question at the centre of many of them was not 'who wrote this' but 'what does this utterance mean, and who initiated the problematic action'.

Entrapment is a legal defence in the United States that requires showing the government induced a crime. Shuy argued that conversational analysis of undercover recordings could show precisely how an agent's language structured an interaction to lead a suspect toward an incriminating statement. The question is not just 'did the suspect say X' but 'under what conversational circumstances did they say it, and was the meaning the prosecution attributes to it the one that a reasonable participant in that conversation would have understood'.

Shuy's books, including Language Crimes (1993) and Fighting over Words (2008), were written explicitly for a practitioner audience, including lawyers with no linguistics background. They made forensic linguistic reasoning accessible in a way that Svartvik's more technical monograph had not, and they helped establish the expectation that a linguistic expert could explain their reasoning to a jury in plain terms.

For much of the field's early history, forensic linguists worked in isolation. A linguist might be instructed as an expert in one high-profile case, publish an analysis, and then return to their main academic work without building a sustained practice. The formation of the International Association of Forensic Linguists in 1993 changed that structural isolation. Founding seminars had been held at the University of Birmingham in 1992, and the association grew from that initiative.

The IAFL brought together academics, consultants, and practitioners from across the English-speaking world and beyond. Its biennial conferences created a venue where case studies could be presented, methods debated, and errors in one jurisdiction flagged for practitioners in another. The association also developed ethical guidelines for expert witness work, including the obligation to give balanced evidence and to state clearly where findings are inconclusive.

The journal Language and Law / Linguagem e Direito (the second title reflecting the journal's bilingual Portuguese co-publication, reflecting significant engagement from Brazilian scholars) provides the field's main peer-reviewed publication venue. The journal publishes casework analyses, theoretical discussions of methodology, and sociolinguistic studies of legal language.

Between 1978 and 1995, Theodore Kaczynski, known as the Unabomber, carried out a mail-bombing campaign in the United States that killed three people and injured twenty-three. The FBI published a 35,000-word manifesto at the bomber's demand in 1995. It was the manifesto that broke the case: Kaczynski's brother David recognised distinctive phrases and arguments from his sibling's letters and alerted the FBI.

The case was solved by family recognition rather than formal forensic analysis, but it drew sustained attention to what systematic linguistic analysis of such a document could have produced. Researchers subsequently studied the manifesto as a methodological benchmark, asking whether computational and corpus-based methods could have identified Kaczynski's writing style from among a large population of comparison texts. The answer was largely yes, but only with a suitable comparison corpus, which in a real investigation would have been unavailable until a suspect was identified.

A striking feature of forensic linguistics's early history is how many of its landmark contributions were retrospective: analysts re-examining cases where a conviction had already been obtained, looking for the linguistic evidence that had been missed or distorted at trial. Evans, the Birmingham Six, Judith Ward, Derek Bentley: these are all cases where a later linguistic analysis contributed to a review of a wrongful conviction.

This pattern had a formative effect on the field's self-understanding. Forensic linguists came to see part of their professional function as protecting against the specific risks of language evidence: police officers who restructure a statement while typing it up, interviewers whose questions shape the answers they get, courts that interpret a phrase in the way the prosecution prefers without asking how ordinary speakers would understand it. The discipline is not merely positive, finding evidence of what happened. It is also protective, identifying where language was used in ways that could mislead a tribunal.

This protective function is one reason the ethical standards the IAFL developed, and the obligation to give balanced evidence rather than advocacy, are taken seriously in the field. The history shows clearly what happens when linguistic evidence is not treated with proper scepticism. Getting that lesson embedded in training rather than learned only from high-profile failures was the institutional project of the IAFL's first decades.