The latest New Yorker has a very informative and thorough piece on forensic linguistics, in the print version and at
The following is my letter to the editor:
As a practicing forensic linguist (since 1979; I have a PhD in linguistics from the University of Chicago and bachelor’s and master’s degrees from Brown, also in linguistics), I thoroughly enjoyed your article on the profession – but with mixed feelings.
It was gratifying to see forensic linguistics, which is not as sexy or yucky as rape kits and maggots but provides valuable information and deserves its own CSI segment, getting long-deserved respect. On the other hand, it is beneficial to forensic linguists that most people continue to give no attention to their writing style and to believe that they don’t have one. Now, if any astute New Yorker now knows that it’s possible to disguise your writing style, he/she may try it, though usually with limited success, because the layperson can’t alter style systematically (e.g., hyphenation patterns).
I rarely accept “disguising his/her writing style” as an explanation. It’s more of a cop-out: the analyst’s failure to uncover the real answer.
So your article, with Jack Hitt’s usual excellent writing, may make people better informed – but it may make my job a little harder.
As regards the role of the human versus the electronic brain, yes, it’s true, the algorithms are human-encoded. And there’s the question of what they actually tell us.
The experienced text analyst can assess the significance of a particular decision (in, say, grammatical structure or word choice). The point I would like the article to have made is that the creation of a text – such as an email written under psychological or time pressure – involves a universe of decisions that are rapidly executed, formulaic (as in the writing of dates and salutations), and often simultaneous, in every aspect of language.
The number and combination of variables is thus quite large, and no two people will make all of the decisions in the same way; the fingerprint analogy is thus accurate. The linguist who is sufficiently sensitive to text can at least identify the broad outlines of the writer’s style. Even (perhaps especially) at the highest levels of language proficiency, there are nuances and differences.
The writer of document A capitalizes Hell, but the writer of document B does not capitalize god. Are they the same person? I doubt it. Someone who capitalizes religious terms will do it consistently; besides, “God” is supposedly the strongest religious concept, far superior to “hell.” Such subtle differences contribute to author identification.
I also wish you had mentioned the many other forensic linguistic sub-disciplines – minority language rights, the language of the courtroom, plagiarism, document interpretation, and many others.
Otherwise, an outstanding article.
Alan M. Perlman
PS. It was good to see Foster again discredited. This guy does not do linguistics, but instead of admitting that, he spun himself a spurious fame.
PPS. I was recently called upon to critique one of Jim Fitzgerald’s analyses. To my amazement, the man mangles basic linguistic terminology. In my report, I wrote “His ineptitude is astonishing.” I have no idea why the FBI finds him so useful.