Gapped text

The letters were rich in evidence of this kind. Their author never used a contraction, wrote ‘to-day’ with a hyphen, as was standard in the first half of the last century, and had an idiosyncratic habit of placing a comma before ‘too’ at the end of a sentence. Most striking of all was the phrase ‘it is not to be wondered at’, which appeared in seven of the twelve letters, usually just before the most wounding accusation.

Her explanation, as it was reported to me later, was almost banal. She had run the club efficiently for twenty years and had been replaced, without much ceremony, by a younger man with a laptop. The letters began a month after the vote. She had not, she insisted, meant anyone to resign; she had simply wanted people to notice how much better things had been run in her day.

What the detective in charge wanted from me, she made clear at our first meeting, was a name. What I could offer was something rather more modest. Forensic linguistics, at least as I practise it, does not identify writers in the way that DNA identifies a suspect. It compares, cautiously, the habits of an unknown text with those of known ones, and states how consistent they are. Television has a great deal to answer for.

Most of us are in the same position. We take enormous care over what we say and hardly any over how we say it, and it is precisely in the ‘how’ that we are most consistently ourselves. A colleague of mine likes to say that people find it far easier to change their minds than their prepositions. It is a joke, but only partly.

Then, in a letter to the parish magazine about the state of the churchyard, I came across ‘it is not to be wondered at’, used once, quite unremarkably, in the middle of a paragraph. The same writer hyphenated ‘to-day’, avoided contractions throughout and, in a separate note about parking, put a comma before a final ‘too’. None of the other eight samples contained more than one of these features.

The breakthrough, when it came, owed little to my reading. A computer program I had been testing, which compares the frequencies of very common words across large numbers of texts, identified sample three as a close match with the letters. Its author proved to be the treasurer’s brother-in-law, who had been quarrelling with him for years over an inheritance.

That turned out to be easier to obtain than I had feared. Village life generates a remarkable quantity of paper, and the police were able to collect minutes of meetings, letters to the parish magazine, Christmas round-robins and, in one case, a complaint to the council about a neighbour’s hedge that ran to four closely argued pages. In all, I had samples from nine people, which I read without being told who had written them.

None of these alternative explanations survived scrutiny. When I finally wrote my report, I said that the writer of sample six was a likely author of the letters and that I had found nothing inconsistent with that conclusion. It is a carefully hedged sentence, and I have learned from experience that the person reading it will usually hear something considerably more definite.