Natural Language Processing :: Computer Technology

Natural Language Processing To digest natural language implies understanding, a function that is uniquely human. To understand something implies to have senses that interpret the world such as emotions and awareness of our own physical experiences. When someone tells a story, we rely upon previous experience for interpretation. We form a reaction, our heart rate may change, we may start sweating, we may relax or tense, and feel certain emotions such as fear. Upon getting new information, a persons attitude may change or the way they think may change. A computer made of metal simply does not have the faculties to experience the world as we do. It can be programmed to respond in such a way that mimics a human response, but can not be considered to be really understanding what it is doing. Recall the story of Helen Keller, how she finally began to learn a language when she was given immediate experiential feedback. The teacher would pour water on her and then do the sign language for the word in her hand. The foun der of Toastmaster's organization started it on the premise that people learn in moments of pleasure, and structured the organization so it would provide just that. A computer would not have the senses to make such understanding of these words and experiences possible. In addition to its lack of cognitive ability, a computer can not form expectations based upon a situation. A political speech can be very serious, but when seen on Saturday Night Live, it will be interpreted as funny. A similar problem is the extra meaning conveyed by the tone of voice or body language. We could always program an exhaustive data bank with all the different possibilities of input, but the system as we know it could not search all of these within a reasonable period of time, nor could it adapt to future changes. One of the problems with natural language processing systems is that humans themselves are often not very good natural language generators or processors. We often apply our own bias and expectations to what we hear. Two people can make eye contact and set up a whole series of background such that they now what each other are talking about. That very fact may be a bonus for computer natural language processing because we can predict with certainty how the system will interpret the information and therefore have greater clarity than people.