Forensic linguistics isn’t psychology

For people interested in social engineering this connection comes up a lot, so let me take some pixels to explain it. Bear with me if you will.

Language is a system. Language seems to share certain features across all our human versions of it. And yet language is infinitely productive. You constantly create sentences so unique that a Google search will probably find your particular expression. That’s insanely cool.

Linguistics is the study of how languages work. For instance, you might not think it but apparently an awful lot of us humans expect determiners in our sentences, including definite ones (not just a dog, but the dog). Some languages use a suffix on the noun to indicate definiteness, but there’s so much expectation in the sentence for a determiner that you can even put in another determiner and indicate definiteness twice. English used to do this, in phrases like “Ye Olde Shoppe”.

That’s pretty far from anything psychological. No one thinks about such language use as it happens, they just do it — or they don’t, if they are natively speaking a language that doesn’t use determiners at all, like Russian or Japanese. You don’t stop to think about whether to say a dog or the dog — you know when to use each phrase depending on what you mean. That’s the definition of native fluency. In fact, what defines language is what its native speakers do with it, no matter what dictionary fans might say.

You know intuitively how language works but linguistics lets you describe what you know. Modal verbs generally describe a contextual universe and the probability of something happening in that universe, for instance. You don’t have to know that to use modal verbs like may, can, should, and might. But you need linguistics to describe what modal verbs do.

And forensic linguistics asks what the language is doing in language evidence (that is, evidence made of language. Lots of evidence isn’t made of language but lots of evidence is). So using FL we can describe better what we intuitively know but might not understand when it has to be put in the context of a legal case. You might already know that a sentence like “I might only have one match, but I can make an explosion” could be a threat (expression about the future), but certainly isn’t a confession (expression about the past). But if you need to explain that in court, you need the linguistic tools to explain what has occurred. (And with linguistic analysis, you might find out your intuitive assumption is wrong.)

Now some linguistic features change with context. Politeness markers appear more often when the speaker is in the immediate vicinity of a stranger — as you might expect. Is that psychological? Not really. One could get into a case study of one particular speaker and whether he spoke as he did because he considered the other people in the room to be strangers or not, but that’s not what linguistics does. It looks at language situations and describes what happens in them. Sometimes there’s enough data to make quantified statements; sometimes there’s only enough to fill out a qualitative analysis. But every language situation produces its own language data.

Adding context to linguistics can push you very far in the direction of interpreting what a speaker thought or meant. There have been enormous miscarriages of justice in court cases where a speaker is assumed to have understood what a court proceeding was and how language works in English, but where those two assumptions were not true. Young people from a culture where speaking truth to authority simply isn’t permitted cannot be assumed to have understood the requirements for truth when testifying in a courtroom situation. In such a case the social context inflects a specific language encounter. But one doesn’t have to know what the speaker was actually thinking to make the case that a teenager trained never to contradict an elder might not answer truthfully when the truthful answer requires contradicting a much older lawyer, speaking in a second language, in a courtroom environment that is totally new to them.

Down at this level every linguist is going to make his or her own decisions. I feel most comfortable with probabilities. In a universe of two suspects, can forensic linguistic analysis differentiate which one probably wrote a purported suicide note? Assuming that it is the case that one of the two of them did, one can assign a high probability of success to such author attribution. Do we know what either suspect was thinking? Is it even the type of question that comes up? No.

There are great books out there about the psychological aspects of social engineering and the art of the short con. It’s a completely different field from forensic linguistics.

There are a bunch of great books out there too on forensic linguistics, if you are interested. Look for books by Roger Shuy, Malcolm Coulthard and Alison Johnson. And you can visit Hofstra’s site on its Institute for Forensic Linguistics, Threat Assessment and Strategic Analysis to find out more about the faculty there including director Dr. Robert Leonard.