Grégory affair: what does stylometry consist of, this technique for analyzing writing which made it possible to relaunch the investigation?
Thirty-six after the fact, the shadow of the raven still hangs over the Gregory affair. Up to now, no expert’s report has made it possible to identify with certainty the author of the letter sent to the Villemin couple on October 16, 1984. This handwritten one-page letter claims the murder of the 4-year-old boy, of whom the body was found in Vologne (Vosges). “The investigations have resumed and are progressing”, indicated Thierry Pocquet du Haut-Jussé, the prosecutor general of Dijon at franceinfo, Wednesday December 16, confirming information from Parisian.
“Hearings have been carried out by the president of the investigative chamber of the Dijon court of appeal in recent weeks”, indicates the magistrate. The investigation was partly relaunched thanks to the expertise of a Swiss company, OrphAnalytics, specializing in stylometry, a discipline which consists in authenticating the author of a document thanks to his personal syntax. In 2017, Claire Barbier, the Dijon investigating judge in charge of the case, had mandated this company to be able to use stylometry and try to identify the raven, revealed Eastern Republican. The magistrate had opted for this foreign company, “in the absence of a known expert in France in this new specialty”, indicated the Lorraine daily.
Stylometry, a technique that focuses on the author’s syntax
To identify the author of a text, OrphAnalytics compares the syntax of a text to the unknown author with that of a text of which the author is known to determine whether the two texts were written by the same person. “The syntactic signal largely predominates along a text because, more stable, it measures the style choices followed by an author”, explains the company on its website.
To do this, OrphAnalytics uses powerful algorithms. Unlike the writing comparison, which relies more on the shape of words and letters, stylometry studies syntax, punctuation or even repetition between words. “By their speed of execution, algorithmic approaches to text authentication are distinguished from linguistic methods which require knowledge of the language, can we read on the company’s website. The algorithms implemented in OrphAnalytics’ delta2T software determine the characteristic profile of a document by systematically identifying the patterns used in and between words, in and between sentences. “
A method little used in France
In France, this technique, which is only based on syntax, is not widespread among writing experts.. “The Swiss use the term to learn more about how to turn sentences. In France, we can identify certain linguistic similarities and this is part of the protocol that we call ‘graphic habits’, explains to franceinfo Christine Navarro, expert in writing and documents, approved by the Paris Court of Appeal and the Court of Cassation. The Swiss are more in the background, thanks to algorithms. For me, this can be complementary to the expertise of writing. “
“Writing is a beautiful science, but there are very few people to develop research [en France]. I was interested in this technique that we called ‘linguistics’ ten years ago while talking to a Belgian police officer. I worked in a lab that did not have the financial or human resources to go into the question in depth. You can’t be an expert in everything … “
For Christine Jouishomme, also a writing expert at the Paris Court of Appeal, stylometry cannot be just one element of study. “It’s a big word for not much. It’s a technique for studying grammatical structure, which is in the realm of learning to write., she told franceinfo. If we only study syntax, that would mean that we are not studying the space between words, the rhythm of the writing or its formatting and we cannot do without it. The syntax can be tricked if someone dictates a text to you, for example. It is very difficult to find the truth by a single item “, she says.
Skeptical defense lawyers
Stéphane Giuranna, lawyer for Marcel Jacob, brother of Monique Villemin, grandmother of little Grégory, said Wednesday on franceinfo that “we do not advance” towards the truth in this matter. “This expertise was entrusted to a Swiss organization in 2017, three years ago, for several tens of thousands of euros, recalls Stéphane Giuranna. The report should have been tabled in 2018. However, we no longer had access to the procedure. This pseudo-expertise is a hoax. The style and syntax of the writings were compared, in the entourage of the Villemin family, with the letters of the raven including that of claim. “
Jacqueline Jacob, wife of Marcel and great-aunt of Grégory Villemin, was designated in 2017 by writing experts as the author of two letters sent to the child’s parents in 1983. “The conclusions are confusing” for her, explained at the time Jean-Jacques Bosc, Attorney General of the Republic of Dijon. Jacqueline and Marcel Jacob had been indicted in 2017, a procedure canceled for a technical defect a year later.
The geographical proximity of the protagonists of this case remains a brake on drawing conclusions on the basis of expertise in stylometry, according to Stephane Giuranna. “They all live in the same place, have attended the same establishments, have the same professional background and the same sociological profile. They all have the same style, the same syntax, they are all born in the same pool. “
Gérard Welzer, lawyer for Marie-Ange Laroche, whose husband, Bernard Laroche, was once suspected of being the murderer of the little boy before being exonerated, also called for caution Wednesday on franceinfo. “There have been so many slippages, so many victims, so much waste, that if there is an incontestable new element, fine. But if there is no incontestable new element and we throw away once again grazing this or that person, we will have made one more mess “. Remember that Bernard Laroche was killed in March 1985 by Jean-Marie Villemin, the child’s father, who suspected him of having killed Grégory.
According to The Parisian, Swiss technique has enabled “to attribute letters of the mysterious raven to a person subjected to the expertise”. During her request for expertise in 2017, Claire Barbier asked OrphAnalytics to analyze the four main letters of the case sent to the Villemin spouses (March 5, April 27, May 17, 1983 as well as the letter of demand of October 16, 1984). For now, the Dijon prosecutor’s office has not yet revealed the name of the author (or author) of these anonymous letters, at the heart of this sprawling investigation.