A Linguist's View on Audience Targeting
In this post, I want to outline a linguist’s take on Semantic Behavioral Targeting. The first question that might come to your mind is “what is a linguist anyway?” No, it is not (necessarily) a polyglot, not (necessarily) a scholar of Greek rhetoric, nor is it (ever) an alternative pronunciation for the well-known Tuscan carbohydrate (you’d be amazed at what people come up with).
A linguist is anyone who approaches natural language (that is, the language people use every day to communicate) as an object of scientific inquiry, using scientific methodologies to test hypotheses about the human capacity to learn, produce and comprehend speech.
I want to outline why a former Ph.D student in linguistics was attracted to a job in an industry he knew nothing about.
After my Ph.D program, a friend referred me to a job at another company in marketing technology. The ever-changing, rapid growth environment of digital advertising was a welcome shift from the glacial pace of academia. But, I also realized that if I stayed in the industry, I would have to say good-bye to the sophisticated, data-processing techniques that I had loved using as a graduate student.
There is no shortage of data in marketing tech. In fact, there is far more data available to the average digital media company than there is to a scientist. But, no one seemed to know what to do with it. Rather than using log data, for example, to understand a user, companies were content to simply label someone a “sports fan” because they had been on a few sports web pages – clearly the only thing computers were being exploited to do was scale, without any real depth.
I then heard about a company that used Semantic Behavioral Targeting – scientifically proven natural language processing techniques (NLP) to truly understand users.
Semantic Behavioral Targeting uses NLP to find the terms that uniquely characterize a user. Regardless of whether this is the right approach (which, I argue below, it is), the first question is if it is even a feasible approach. As a linguist, I know it is. Language data is rich enough to give you a sufficiently unique view of the user, and is sufficiently easy to process.
Quantifying and processing visual information, such as on social platforms, is simply too difficult to scale. And, it’s not precise enough. You and I may each see a picture of people on the beach, but while you think of a vacation, I might think of the danger of sharks. The message is ambiguous.
Location data, while easily quantified and unambiguous, simply does not scale. As online shopping grows, location data will also tell less us about consumers. Language data is unambiguous, and can be processed at vast scales. And, most importantly, NLP techniques make sense.
Finally, there is the philosophical argument underpinning why a linguist was excited that a marketing company would embrace language: language is literally the stuff of thought, as Steven Pinker argues in his book of the same title. The primary function of language is not to facilitate communication (that’s its secondary function). The primary function is to facilitate consciousness. It’s what separates us from other animals. Language is the medium by which the brain processes information. So, to understand people’s language is to understand people’s thoughts. And, to understand people’s thoughts is to understand the consumer.