Language

Google Translate goes AI

Wednesday, 28 September, 2016 0 Comments

From now on, Google Translate will rely more on AI (artificial intelligence) when it translates languages. Alphabet, the parent company, claims that its brand new Google Neural Machine Translation system will reduce errors by 80 percent compared to its current method.

Google Translate Until today, Google has used what is called “phrase-based translation,” which is standard for the industry. With this method, a hand-coded algorithm breaks down a sentence into words or phrases and tries to match them a vast dictionary. The new system will use that same large dictionary to train two neural networks, one of which will deconstruct the original sentence to figure out what it means, while the other generates text in the output language.

Because AI algorithms don’t rely on human logic, they can often find better ways to do the job compared to the hand-coded algorithms, say the engineers. And as the network learns how to translate, no longer spending time dividing sentences into words or phrases, it discards the rules that humans thought were best and concentrates fully on the outcome. Such is the nature of AI. As Alan Turing wrote in 1950: “I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted.” (Computing machinery and intelligence). We’re getting there.

Google is releasing its new translation system for Mandarin Chinese first, and then adding new languages over coming months.


Hodge and his lexicographer

Sunday, 18 September, 2016 0 Comments

On this day in 1709, Samuel Johnson was born. The poet, essayist, moralist, literary critic, biographer, editor and lexicographer spent nine years writing his Dictionary of the English Language, which was published in 1755 and continues to enlighten and amuse: “Oats: a grain which in England is generally given to horses, but in Scotland supports the people.”

The drudgery of lexicography was alleviated somewhat by Hodge, a cat the good doctor loved, and his friend and biographer James Boswell found Johnson’s relationship with Hodge so important that he preserved it for posterity:

“I never shall forget the indulgence with which he treated Hodge, his cat; for whom he himself used to go out and buy oysters, lest the servants, having that trouble, should take dislike to the poor creature. I am, unluckily one of those who have an antipathy to a cat, so that I am uneasy when in the room with one; and I own, I frequently suffered a good deal from the presences of this same Hodge. I recollect him one day scrambling up Dr. Johnson’s breast, apparently with much satisfaction, while my friend smiling and half-whistling, rubbed down his back, and pulled him by the tail; and when I observed he was a fine cat, saying why, yes, Sir, but I have had cats whom I liked better than this; and then as if perceiving Hodge to be out of countenance, adding, but he is a very fine cat, a very fine cat indeed.”

Hodge is remembered by a bronze statue outside the house at 17 Gough Square in London he shared with Johnson and Barber, Johnson’s black manservant and heir. The statue shows the cat sitting next to a pair of empty oyster shells atop a copy of Johnson’s dictionary, with the inscription “a very fine cat indeed”.

Hodge


Yes, he did.

Friday, 2 September, 2016 0 Comments

On Wednesday, here, our post was about the Irish poet Seamus Heaney, who died on 30 August 2013. As Henry Miller put it in Tropic of Cancer: “In this chthonian world the only thing of importance is orthography and punctuation. It doesn’t matter what the nature of the calamity is, only whether it is spelled right.”

He did


Speech II: Tom Wolfe vs. Noam Chomsky

Wednesday, 17 August, 2016 1 Comment

Tom Wolfe’s new book, The Kingdom of Speech, looks at the work of four major figures in the history of evolution and language: Charles Darwin, Alfred Russel Wallace, Noam Chomsky and Daniel Everett. A 15-page excerpt appeared in the August issue of Harper’s Magazine titled “The Origins of Speech: In the beginning was Chomsky” and it focused on rise of Chomsky and Everett’s challenge to Chomskyism within the world of linguistics. The story begins in 1957, when Chomsky was 28. He wrote a book “with the opaque title” Syntactic Structures that turned the world of linguistics “upside down,” writes Wolfe. Snippet:

Language was not something you learned. You were born with a built-in “language organ.” It is functioning the moment you come into the world, just the way your heart and your kidneys are already pumping and filtering and excreting away.

To Chomsky, it didn’t matter what a child’s first language was. Whatever it was, every child’s language organ could use the “deep structure, ” “universal grammar, ” and “language acquisition device” he was born with to express what he had to say, no matter whether it came out of his mouth in English or Urdu or Nagamese. That was why — as Chomsky said repeatedly — children started speaking so early in life… and so correctly in terms of grammar. They were born with the language organ in place and the power ON. By the age of two, usually, they could speak in whole sentences and generate completely original ones. The “organ”… the “deep structure”… the “universal grammar’… the “device” — as Chomsky explained it, the system was physical, empirical, organic, biological. The power of the language organ sent the universal grammar coursing through the deep structure’s lingual ducts to provide nutrition for the LAD, which everybody in the field now knew referred to the “language acquisition device” Chomsky had discovered.

Harpers Along with Chomsky’s linguistics, Tom Wolfe devotes a great deal of space to Chomsky’s politics, which have grown increasingly bizarre over the years, to the point where he ascribes almost all evils in the world to the USA. Despite, or possibly because of, such derangement, he remains a darling of the left-liberal media and nothing he says, no matter how absurd, is taken seriously by his credulous disciples.

Then, OOOF!

Typically Wolfe, the capital letters are introduced to make a big point and the biggest one concerns a 13,000-word article that appeared the August–October 2005 issue of Current Anthropology entitled “Cultural Constraints on Grammar and Cognition in Pirahã” by Daniel Everett. Pirahã is a language spoken by several hundred members of a hunter-gatherer tribe in the vast Amazon basin and it does not contain any recursion, which is central to Chomsky’s theories, and “it was the Pirahã’s own distinctive culture, their unique ways of living, that shaped the language — not any ‘language organ,’ not any ‘universal grammar’ or ‘deep structure’ or ‘language acquisition device’ that Chomsky said all languages had in common,” declares Wolfe.

Piraha

The Pirahã sentence ‘There is a paca there’ uses just two words: káixihíxao-xaagá, meaning, paca exist there

Tom Wolfe is very enjoyable on the academic skulduggery used by the Chomskyians to denigrate Everett’s work and destroy his career, and one is left with the impression that the professorial class is filled with characters similar to the consigliere and caporegime of the Mafia. Filled with loathing for Chomskyism, Wolfe concludes thus:

“In three decades nobody had turned up any hard evidence to support Chomsky’s conviction that every person is born with an innate, gene-driven power of speech with the motor running. But so what? Chomsky had made the most ambitious attempt since Aristotle’s in 350 B.C. to explain what exactly language is. And no one else in human history had come even close. It was dazzling in its own flailing way — this age old, unending, utter, ultimate, universal display of ignorance concerning man’s most important single gift.”

OOOF!


Speech I: Oliver Kamm vs. Tom Wolfe

Tuesday, 16 August, 2016 0 Comments

The Kingdom of Speech is published by Jonathan Cape but might as well have been issued by the Jehovah’s Witnesses, whose tracts at least have the merit of being funnier.” So writes Oliver Kamm in a devastating put-down of the latest book by Tom Wolfe. Along with being the author of Accidence Will Happen: The Non-Pedantic Guide to English Usage, Kamm is a leader writer and columnist for The Times and it was in that newspaper on Saturday that he took Wolfe to the reviewing woodshed.

The Kingdom of Speech Tom Wolfe argues that speech, not evolution, sets humans apart from animals and is responsible for all of our great achievements. He targets Charles Darwin and Noam Chomsky in The Kingdom of Speech when arguing that there is no evolutionary explanation for language, particularly abstract language. Kamm differs, however: “Wolfe’s theory that words are a memory aid — a mnemonic system — likewise falls apart on a moment’s reflection. Words like cat and dog and run and jump might help us to remember things in the world, but what about words like not and very and whether and however?”

Readers who purchase The Kingdom of Speech in the hope of acquiring an invigorating clarification of the big ideas at the heart of the debate about morphology, syntax, phonetics and semantics will be misled, claims Kamm, who shows no mercy in his critique:

“It’s a celebration of ignorance: a vain, sneering and calumnious piece of fluff in which Wolfe misunderstands his subject and misrepresents leading thinkers, notably Darwin and the linguist Noam Chomsky. It’s not even stylishly written. What I learnt from it is that a crotchety celebrity of vaulting hubris and small mind doesn’t feel constrained by canons of evidence and accuracy.”

Tomorrow, here, Wolfe attacks Chomsky.


Girls gone, gone, gone, gone, gone

Monday, 8 August, 2016 0 Comments

The huge commercial success of both Gone Girl by Gillian Flynn and The Girl on the Train by Paula Hawkins has not passed unnoticed by the modern version of Grub Street, the London thoroughfare once famed for “its concentration of impoverished ‘hack writers’, aspiring poets, and low-end publishers and booksellers.” Young women have been pressed into service by this cruel trade in the hope that another lady-vanishes winner can be typed while the genre is hot. Five hopefuls for the crock of gold:

Good as Gone by Amy Gentry. “When 13-year-old Julie Whitaker is kidnapped from her bedroom in the middle of the night, her family shatters. As the years go by and the search for Julie turns up nothing, even her mother Anna begins to lose hope. Then one night, the doorbell rings.”

All The Missing Girls by Megan Miranda. “It’s been ten years since Nicolette Farrell left her rural hometown after her best friend, Corinne, disappeared from Cooley Ridge without a trace.”

Missing, Presumed by Susie Steiner. “As soon as Detective Sergeant Manon Bradshaw sees the photograph of missing Edith Hind — a beautiful Cambridge post-grad from a well-connected family — she knows the case will be big.”

Under The Harrow by Flynn Berry. “A woman is missing in the East Riding. She vanished from Hedon, near where we grew up. When Rachel will learn of the disappearance, she will think it’s him.”

You Will Know Me by Megan Abbott. “This would be the piece that mattered most later, months later when Ryan was gone. She would think of their arrival and wonder why she hadn’t seen it all coming. But who could have seen anything at all that night but their bright-spangled beauty?”

According to the Paris Review, George Gissing’s New Grub Street (1891) “is good for a whole host of reasons, but it’s a particularly potent corrective to the current cottage industry centering on ‘the writing life’ — in which literary production is seen as glamorous, in which photos of writers’ desks appear on Pinterest and readers obsess over the perfect pen with which to write their buried masterpiece.” Ouch!

“Literature nowadays is a trade. Putting aside men of genius, who may succeed by mere cosmic force, your successful man of letters is your skilful tradesman. He thinks first and foremost of the markets; when one kind of goods begins to go off slackly, he is ready with something new and appetising. He knows perfectly all the possible sources of income. Whatever he has to sell, he’ll get payment for it from all sorts of various quarters; none of your unpractical selling for a lump sum to a middleman who will make six distinct profits.” — George Gissing, New Grub Street


Future sex with gynoids and guynoids

Wednesday, 3 August, 2016 0 Comments

The word “gynoid” was used by Gwyneth Jones in her 1985 novel Divine Endurance to describe a female robot slave character in a futuristic China. Does this mean, then, that the male equivalent is a “guynoid”? Not quite. Gynoid is created from the Ancient Greek prefix gyno– (of or pertaining to women or the female reproductive system) + android, a Greek word used to refer to robotic humanoids regardless of gender. However, the Greek prefix “andr-” means man in the masculine sense and because of this android is used to describe male-styled robots. Given the established etymology, it’s going to be a battle to replace androids with guynoids.

All this is by way of saying that sex with robots is very much in the news. Let’s take three of today’s headlines, starting with The New Scientist. “Could sex robots and virtual reality treat paedophilia?The Daily Mirror is more of a mass-market publication: “Expert to publish ‘how to build your own sex robot’ handbook after Scarlett Johansson lookalike success,” while The South China Morning Post brings us back to the gynoid world of Gwyneth Jones: “Sex and robots: How mechanical dolls may press all the right buttons for lonesome guys.”

Actually, that last headline is quite topical in light of the work being done by Kathleen Richardson, a Senior Research Fellow in the Ethics of Robotics at De Montfort University in Leicester. Last September, she published a position paper titled “The Asymmetrical ‘Relationship’: Parallels Between Prostitution and the Development of Sex Robots.” Snippet:

“Following in the footsteps of ethical robot campaigns, I propose to launch a campaign against sex robots, so that issues in prostitution can be discussed more widely in the field of robotics. I have to tried to show how human lifeworlds of gender and sexuality are inflected in making of sex robots, and that these robots will contribute to gendered inequalities found in the sex industry.”

The debate about the gendering of robots and the sexualized personification of machines is on.

Ex Machina


The Robolution federator

Tuesday, 2 August, 2016 0 Comments

The Fourth Industrial Revolution’s upgrading of English vocabulary is a regular theme here and the prospect of public presentations on the subject in October and November is concentrating the mind, to paraphrase Dr Johnson. We’ve had some gems recently and more are to come. Central to the revolutionary stuff going on right now is robotics.

Definition: “Robotics is the branch of mechanical engineering, electrical engineering and computer science that deals with the design, construction, operation and application of robots, as well as the computer systems for their control, feedback and data processing.”

If you create an €80 million private equity fund dedicated to robotics, you’re going to need a name for the venture; one that combines the essence of the business with its revolutionary role in 21st-century industry, ideally. Robolution The result is… Robolution. Or, more precisely, Robolution Capital. But there’s something slightly unmelodious about the word “Robolution,” with its hints of ablution and absolution. Sure, it’s an attempt to capture an element of “revolution,” but the “robo” bit at the front doesn’t quite make a harmonius unit, does it? Perhaps it sounds better in French because Robolution Capital is based in Paris.

Along with robotics, Robolution Capital is focussing on artificial intelligence (AI) and the Internet of Things (IoT), two very hot areas right now, and this is why it defines itself as a facilitator, an accelerator and “a federator at the heart of the ecosystem of entrepreneurs, corporates, public organizations, universities and research centers.” What’s a federator? The usually indefatigable Wiktionary does not have an entry for the word and Techopedia offers “Federation” from the world of enterprise architecture that allows interoperability. The word, however, is a version of fédérateur, the French noun that means “unifier.” And with its philosophy and its focus on robotics, AI and the IoT, Robolution is true federator.

News: 360 Capital Partners, an early-stage VC business based in Milan and Paris has just done a deal with Orkos Capital, also based in Paris, to manage Robolution Capital.


WoTD: Servitization

Monday, 1 August, 2016 0 Comments

“I spent the past week at the University of Birmingham in England with a group of 16 Operations and Economics Professors from across Europe,” wrote Rosemary Coates on 6 July in Supply Chain Management Review. She was there to lecture and to represent the Reshoring Institute, which provides “research and support for companies bringing manufacturing back to America.” As we know, manufacturing jobs will be one of the hottest topics in the so-called Rust Belt states during the US presidential campaign, and both candidates have made their positions on the subject clear.

In her blogpost, Ms Coates noted, “Some of the biggest buzz of the week was around the idea of Industry 4.0 (the Internet of Things) and Servitization.” What might appear to some as a misspelling there, “servitization,” is a real word. But what is it?

“This is the process of companies transforming from simply producing a product to including service in the total product offering. The complete product package includes field service, service level agreements and pricing for spares and replacement parts. European manufacturers are way ahead in Servitization.

Some American companies such as Cisco Systems have been including product services and consulting services in their product offerings for many years. But US companies like Cisco, that understand a fully integrated product offering and co-sell product and services, are few and far between.”

The etymology here involves creating a word from “service + -ization.” One assumes “serviceization” was considered unspellable and so we got “servitization” instead. In jargon-speak, “servitization is a transformation journey that involves firms developing the capabilities to provide solutions that supplement their traditional offerings.”


Language acquisition à la Goldman Sachs

Sunday, 31 July, 2016 0 Comments

The noun is returnship (plural returnships) and it’s a blend of return + internship. Definition: “A returnship is an internship-like program for experienced workers seeking to re-enter the workforce after an extended absence, often in a new line of work.”

The notion of “returnship” is central to The Return Hub, recently launched by Dominie Moss, who has spent 20 years in London’s financial services industry as a commodities trader and then in executive search. According to the “mission statement,” The Return Hub is “a campaign to raise the profile of returning women with employers in the financial sector.” Naturally, it’s got a hashtag: #timetoreturn

If one scrolls to the end of the extensive homepage, this appears: “*Returnship — a term trademarked by Goldman Sachs.” And it’s a fact. Goldman Sachs started its returnship program in 2008 and trademarked the term. “We are committed to help facilitate the ‘on-ramping’ process” is how Goldman Sachs puts it in “Start Your Journey Back to Work with the Goldman Sachs Returnship® Program.”

What’s the real meaning of that ® symbol there? Well, it indicates that a trademark has been federally registered with the U.S. Patent and Trademark Office, which defines a trademark as a “word, phrase, symbol or design, or a combination of words, phrases, symbols or designs, that identifies and distinguishes the source of goods of one party from those of others.”

Note: Trademarks protect the words and symbols that identify the sources of goods and services. Patents, on the other hand, protect inventions and improvements to inventions, while copyright protect artistic or literary works. Unlike copyrights and patents, trademarks can be perpetual, as long as a company keeps using its trademark. The only way to lose a trademark is if it becomes the generic name for a product or service. A trademark does not mean, however, that no one else can use your word, phrase, or symbol in connection with any and all goods and services. It means only that somebody else can’t use a similar trademark with similar goods or services.

One imagines that our learned friends in the City considered this matter in detail before advising The Return Hub to embellish its offering with the applicable asterisked notice at the bottom of the page: “*Returnship — a term trademarked by Goldman Sachs.”


Buongiorno! Amazon’s wake words in Italy

Wednesday, 27 July, 2016 0 Comments

Amazon apre un nuovo centro di sviluppo per l’intelligenza artificiale e il Machine Learning a Torino. That was the welcome news for Italy’s battered economy earlier this week. Translation: “Amazon to open a new artificial intelligence and machine learning development centre in Turin.” The charming capital of Piedmont will soon be home to a batch of software engineers and linguists developing machine learning capabilities for Alexa, Amazon’s cloud-based data and analytics service. This sentence in the press release stood out:

“Alexa usa l’apprendimento automatico in campi come il rilevamento delle parole di attivazione, il riconoscimento vocale basato sul cloud e la comprensione del linguaggio naturale.”

Question: How does one translate parole di attivazione? The available online Italian-English dictionaries are not up to the job and Google Translate offers “words activation” as its best shot. Close, but no cigar. In fact, parole di attivazione are “wake words”. Eh?

Amazon Echo To understand the function of wake words, get an Amazon Echo. This hands-free speaker connects to the Alexa Voice Service to play music, provide news, sports scores and weather forecasts. When you want to use your Echo, speak the word “Alexa” and the device comes to life instantly. That’s the “wake word”. If you have more than one Echo, you can set a different wake word for each. You can pick “Amazon” or “Echo” as the wake word. And that’s it. Why the paucity of wake words? Well, according to Veton Kepuska, author of Wake-Up-Word Speech Recognition, the challenge is to:

“Detect a single word or phrase when spoken in an alerting context, while rejecting all other words, phrases, sounds, noises and other acoustic events with virtually 100% accuracy including the same word or phrase of interest spoken in a non-alerting (i.e. referential) context.”

See the problem? In its search for usable wake words, Alexa needs ones that are not only easy to pronounce and remember, but are also rare enough that they’re not even used at the start of sentences. Very tricky. As things stand, it’s doubtful Echo owners will be able to choose their own wake word for a long, long time to come. The best hope of the Turin project is that the team there will create an expanded list of words that are unlikely to lead to too many false wakes. No false dawns. No hurry, in other words.

Turin is an ideal location for this venture. It’s the home of the slow food movement.