Print

Surviving Machine Translation - future view: C-3PO & Human Translator

Written by Paul Sulzberger on 11 June 2012.

c3poGoogleRemember C-3PO? He was a protocol droid designed to serve humans and known to be fluent "in over six million forms of communication". His main function was to ensure that meetings of different cultures from his galaxy ran smoothly [1].

Poor C-3PO. Given that there are some hundred billion galaxies in the known universe, a knowledge of the local languages of his own galaxy alone made him quite inadequate for full-strength intergalactic translation...

Nevertheless, if we trace his development back through the millennia to his earliest primitive ancestor, Google Translate, we can get a sense that however limited his grasp of intergalactic languages may have been, C-3PO's technology had come a long way since the early 21st century.

Back then people were still wondering how long it would be before simple droids would put human translators out of work. In those days people were still worried that if English ended up as a lingua franca of planet Earth, the sort of work done by professional translators would largely become redundant, and their livelihoods would vanish.

At that time there seemed to be many threats to an ancient profession which had been central to the very development of civilisation and the dissemination of knowledge across the planet.

WHEN MACHINES BEGAN TO REASON...

2011 saw the birth of new generations of cognitive computers - such as the chips developed by IBM which began to learn through their experiences forming their own theories about what those experiences meant [2]. This, perhaps, was the beginning of the process of getting computers to reason instead of reacting solely based on data that had been pre-programmed.

A futurist of the time, Ray Kurzweil suggested that computers would rival human intelligence by around 2029 [3]. The early 21st century was a time when people began to reason that if the intelligence of a machine "can successfully perform any intellectual task that a human being can" [4], then droids like C-3PO would most certainly replace human translators.

However, the 21st century didn't see an army of C-3POs replacing the planet's human knowledge workers. As machines continued to liberate them from routine tasks, humans took on increasingly sophisticated intellectual tasks.

Bee  AnnaIn any event, it was estimated that the computers of that era had as much processing power and speed as the brain of a bee [4]. Now bees are actually very smart, and have brains composed of some 960,000 neurons. While they developed some sophisticated communication systems, the evolutionary path of the bee never led to the development of language as we understand it - let alone the ability to undertake intellectual tasks of the magnitude of translating Anna Karenina into Finnish or Swahili - feats requiring the complexity of a human brain equipped with an additional 20,000,000,000 neurons. [5]

Despite its limited capability, 21st century technology made some fantastic leaps in the development of machine translation of human languages after IBM's first primitive attempts began in the middle of the 20th century. The appearance of Google Translate in 2006 was one of them. It profoundly changed public consciousness about translation and provided access to an otherwise hidden multi-lingual world for millions of people. By 2012, machines were producing as much translation in a single day as the entire human workforce produced in a single day. [6]

But even as language-capable machines were beginning to develop, 21st century linguistics was only just scratching the surface of how complex human languages systems really were. For thousands of years, linguists and translators had intimations of this complexity, as they struggled to reproduce ideas conceived via the thinking processes of one language in another. Somehow they managed - often by intuition, more often by sheer hard work and skill, but sometimes by simple good fortune.

While 21st century technology eventually mastered simple information transfer between languages, when the transfer of ideas involved human emotions, empathy, persuasion, the resolution of ambiguity and the dissemination of ideas across different cultural traditions - human translation continued to rule supreme.

THE PRECEDENTS OF SURVIVAL

There had been plenty of precedents of how threats to the livelihood of humans from new technology had been resolved. Generations of skilled tradesmen completed long years of apprenticeship to acquire sufficient knowledge to complete such tasks as laying out the pages of books. Typography and typesetting demanded the development of great aesthetic sensitivity and deep technical skill. But in 1985, the introduction of Apple's LaserWriter printer, and the introduction of desktop publishing software blew the industry apart. [7]

Overnight, the skills of typographers and typesetters, honed over a lifetime's practical experience, were available to anyone at the touch of a button - all encoded in a piece of software. Thousands of typesetting firms went broke virtually overnight.

There was a brief moment when the world was flooded with the most atrocious typesetting imaginable, all produced by inexperienced amateurs who lacked sensitivity to type or aesthetic design - skills which can only be acquired from long experience. The new sophisticated tools in the hands of such amateurs did not represent much of leap forward.

SOPHISTICATED TOOLS ... IN THE HANDS OF EXPERTS

However, sophisticated tools in the hands of experts saw an immediate explosion of creativity in typography and graphic design. Who, in 1985, would have thought that the typesetter's craft would mutate into dozens of completely new and conceptually different forms - digital imaging, digital interaction design, web page design, user interface design, usability engineering and user experience design? Liberated from the mundane by their machines, graphic designers and innovative typographers reached for new creative heights.

Human translators too survived the entry of machines into their profession. Just as previous generations had abandoned their quills for the typewriter, which itself soon mutated into the computer, they quickly adapted to printed dictionaries, spell checkers, digitally distributed terminology databases and machine translation. All these new tools progressively unshackled translators from the tedious and allowed them to tackle the increasingly sophisticated intellectual demands of their age - tasks undreamed of at the dawn of the 21st century.

While we have achieved much throughout the eons, new problems in intergalactic translation continue to emerge. Human translators, equipped with the latest bio-quantum computing tools, continue to rise to the challenge of the linguistic difficulties of our age.

The Author

Paul Sulzberger was born in New Zealand and studied second language teaching methodology at Moscow State University in the late 1960’s. He completed his MA (Hons) degrees in modern languages and political science at Victoria University, Wellington.  Later in life he completed his PhD in Applied Linguistics.

Notes:

[1] C-3PO, Wikipedia: http://en.wikipedia.org/wiki/C-3PO

[2] New IBM computer chip mimics the human brain, CNN, http://bit.ly/qA42bN

[3] Kurzweil Is Confident Machines Will Pass Turing Test by 2029 : http://bit.ly/eebdhT

[4] Stong AI (Artificial Intelligence), Wikipedia, http://en.wikipedia.org/wiki/Strong_AI

[5] List of animals by number of neurons, Wikipedia, http://bit.ly/cSpfaF

[6] Breaking down the language barrier—six years in. http://bit.ly/IpSVT0

[7] Desktop publishing, Wikipedia, http://en.wikipedia.org/wiki/Desktop_publishing

Dana Translation Facebook Page   Dana Translation Google+ Page   Dana Translation LinkedIn Page   Dana Translation RSS Feed   @DanaTranslation

Dana Translation

10 Brackendale, Gloucester Road

Kingston Upon Thames, KT1 3RL

United Kingdom