We’ve introduced the notion in the previous article that the brand of Linguistics devised and practiced by Chomsky at MIT and in other prestigious US campuses is indeed a science and not something more akin to say, literary criticism, in and of itself a highly worthy pursuit, but not a scientific one. Science purports to investigate the laws and principles underlying the manifestations of physical phenomena. Entranced by the magical hold it has on human experience, many forget that language in its most direct manifestation is part of the material world.
Phonetics, probably the oldest and most advanced branch of Linguistics, is concerned with the classification and mechanics of production of all human language sounds. Vast libraries exist in secure vaults with recordings of sound patterns that range from the most common such as [p] to the most exotic and rare like African clicks. The sheer range of linguistic sounds pronounceable by the human speaking apparatus is enough to challenge the boldest of intellects. This, however, did not deter characters like Professor Higgins, who managed to teach a poor girl selling flowers on a street corner how to enunciate her words like a fashionable lady of society. How did he do it? How did the genial professor pull it off? Was it all George Bernard Shaw’s fancy?
It is told that Shaw’s character was based on a flesh and blood linguist who could tell on which city block in London people lived just by listening to them speak. To some, this may sound unbelievable, illusionism, parlor trickery. It turns out that it is possible to emulate such feat by employing the same tried and true scientific tools of research, data gathering, and analysis that have been the mainstay of rational inquiry at least since Galileo and beyond.
Like everything in nature, language does not exist in a vacuum but is rather part of an experiential continuum. Years are spent gathering phonetic data in a particular geographical area and producing taxonomies, charts detailing the physically detected features and qualities of phones (vowels and consonants) produced by recorded subjects. These recordings are also used to create phonograms, special wavelike diagrams that visually reproduce the vibrations of discrete phone clusters within a speech stream. Once a delimited linguistic area has been thoroughly investigated, a phonetician will be able to identify its speakers by their idiosyncratic speech patterns just as the trained ear of a virtuoso picks up the slightest tonal variations in a musical piece.
Modern speech recognition software stems directly out of this kind of work. When you make a call and get a “machine” that claims you can talk to it, it’s not quite true. There is no impersonal sentient entity across the ethereal waves who can’t wait to direct you where you want to go. What you have is an automated set of programs: one breaks down the sounds produced by your voice following established vibratory ranges, another puts them back together according to general word formation rules, yet another compares the just reconstructed words to a lexicon of possible sentences regarding the matter you are calling about. Once the sentence has been identified, a specific subroutine picks one out of an array of digitally prerecorded responses and shoots it back to you. The process goes on until you start screaming that you want to speak to an agent!
This exquisite kind of aggravation would not have been possible without the assiduous work of scores of dedicated linguists working at Bell Laboratories and other similar places who methodically broke down a part of the purely physical part of English and transformed this painstakingly gathered data into digitally powered algorithms. Add a sprinkle of context driven procedures and voila, the illusion is served. Smoke and mirrors based on sound scientific truth.