Hammer and Nails

The favorite topic of translators is rates, followed, in the last few years, by machine translation.

On both topics, nearly every conversation ends up in shouting matches, partisan vitriol, and ideological food fights, even in virtual communities. In the past, the same happened with languages and their relative importance.

Il presente articolo è stato estratto dal blog "The Big Wave" , una rivista trimestrale dedicata alle tecnologie linguistiche, pubblicata da sQuid.

---

Commodities are the stuff which goes to make stuff.
John Lanchester, Whoops!


The favorite topic of translators is rates, followed, in the last few years, by machine translation.

On both topics, nearly every conversation ends up in shouting matches, partisan vitriol, and ideological food fights, even in virtual communities. In the past, the same happened with languages and their relative importance.

Rates and machine translation are governed by elements that are beyond the understanding of most people, who confront them on the basis of convictions and passions, sometimes unjustified prejudices, rather than a factual body of knowledge according to a rational approach.

Economists, for example, are used to say that free markets allocate goods efficiently, and that, by allowing people to make mutually advantageous trades, markets allocate goods to those who value them most highly. But in doing so, markets also express and promote certain attitudes towards the goods being exchanged, and this leads to the acknowledgement of the ethic of market based on “getting what you pay for”.

When certain goods are bought and sold, they become commodities, and the willingness to pay for a good does not show who values it most highly, because market prices also reflect the ability to pay. In fact, one of the core principles of economics is the price effect: when prices go up, people buy less of a good, and when prices go down, they buy more.

Economists assume that people decide what to buy by weighing the costs and benefits of the available options , and choosing the one they believe will give them the greatest utility. This also explains why the price system allocates goods according to people’s preferences, without assessing them.

In this respect, information plays a central role, and markets use signaling to overcome information asymmetries. A signaling equilibrium is necessary, though, under which information can be exchanged in honestly and trustfully. In the translation industry, rates — or fees, or prices — are typical signals, and a third party mediating between the buyer and the producer can retain pieces of information and break the equilibrium.

There’s no perfect competition in the translation industry: not all agents are rational and have perfect information; not all buyers can choose the best products, and the market will consequently reward the best buyers with lowest prices, rather than the best producers with higher sales.

For a perfectly competitive market, product prices and quality must be known to all buyers and vendors, who must be infinite, willing and able to buy and supply the product at a certain price.

In addition, buyers and vendors shouldn’t sustain any costs in making an exchange, and shouldn’t find any barriers in entering and exiting the market.

The first two conditions are definitely missing in the translation market. The latter does not suffice, while two more should exist that could severely hamper the market’s efficiency: product homogeneity, i.e. invariance across suppliers, and well-defined property rights determining what may be sold and the rights conferred on the buyer.

This passage from a recent New York Times editorial by Lawrence Summers, former president of Harvard University and former secretary of the Treasury, should therefore come as no surprise:

“English’s emergence as the global language, along with the rapid progress in machine translation and the fragmentation of languages spoken around the world, make it less clear that the substantial investment necessary to speak a foreign tongue is universally worthwhile.”

On the other hand, in commenting these words, David Bellos argued:

“English speakers are now a relatively small minority among global users of English. However, the majority of users of English do not write or speak “English” but, rather, a vehicular tongue called I.E. (International English) which is highly inflected by the (very diverse) native languages of its users.
All users of I.E. are fluent/native speakers of another language, which they can (and do) use for more important, personal, or subtle kinds of communication (such as inevitably arise in business, diplomacy, and treating the sick). The only people disenfranchised by recourse to I.E. are those native speakers of English who do not have another idiom for side-channel consultation.”

Should this be true, however, most of what is really important and worth translating, at least from a business point of view, could be machine translated, just like some suggest.

Also, for perfect competition, no player should be big enough to have the market power to set the price of a homogeneous product. This is not the case of the translation industry, where the largest players together roughly control 20% of the market, and yet have the ability to set the price.

As this wasn’t enough, according to a new Common Sense Advisory report, the leading languages online are those of major industrial nations.

In 2012, 13 languages address 90% of online audiences. The online English-speaking population alone, though, still sums up to more than 60%, with a much higher combined economic buying potential. The Norwegian population is deemed to be 100% online, and yet it is less than 0,4% of the total, even though with a substantial economic buying potential. The Spanish-speaking population represents more than 5%, and Chinese-speaking population 36%, and yet they have no comparable economic buying potential. India will launch its Mars Orbiter Mission in November 2013, its fast-growing economy will be worth US$2 trillion by the end of that year, and yet 40% of its children are malnourished.

This means that market fragmentation will continue in the near future, even more than in the past, and that willingness to pay does not equal the expenditure capacity/spending power.

Machine translation, and translation technology as a whole, would help abridge this divide, and translators should not be afraid of it. On the other hand, whether being scared of something or not, it is better not to cry against it, but rather ignore it — or at least pretend to — and exploit it, proving to be better and stronger.

If it’s true that the probability of events can be measured in terms of the bets people are willing to make, machine translation is here to stay. On the other hand, Jost Zetsche is perfectly right in maintaining that machine translation is a completely different activity than the one performed by translators, and that:

“Human translation, like any human activity, is not guaranteed to be successful, but with the right processes in place, the likelihood of success is extremely high. Contrast that with translation by a computer: if the MT system works well, the likelihood of success is exactly as great as the mathematical probability that the computer can apply.”

Translators have been victims or followers of the law of the instrument: traditionally, they have always been taught to use a very limited set of tools, and rather than being led to lateral thinking to overcome this limitation, they have been indulging on it. This also explains why all major innovations in the translation industry have come from outside. Rather than using their knowledge to analyze the events in the industry, understand them, fight them or take advantage of them, translators have given in uncritically to their convictions. Unfortunately, when too many people believe too deeply, too stridently, in their own convictions, they end up wriggling to impose them on everyone else.

Maybe this explains why outsiders fail to understand the translation industry. Its messages need a lot of translation, and again it should come as no surprise how few and yet untranslated research papers are released, and how many mediocre and sterile articles, essays, and books are published, especially by those who have the opportunity to bring their voice where it can be heard.

Some of these are critical while insightful and yet inconclusive, others have the ambition to trace the recent history of the industry, by offering analysis and forecasting, but they end up only expressing a late and backward hope —if not a helpless daydream.

A reflection on the translation profession: over the years, translation has become increasingly simpler and easier in the eyes of buyers. In fact, the variety and availability of technological tools and their apparent simplicity have been fueling the belief that translation is easily accessible and virtually painless, devaluing the effort and dedication required to acquire the necessary specific skills.

Thirty years ago, even a simple search could take a long time in a library; a translation had to be produced on paper, possibly with a typewriter, and errors, even typos, were a nightmare, so that liquid paper was welcomed as a miraculous invention. In addition, there were very few and too often inadequate paper dictionaries, and pages and pages piled up in glossaries (the real wealth of a translator at that time). Today, a complex search can still take some time, but, in general, search engines allow for browsing and checking a huge variety of sources, while the Internet makes term bases relatively easy to find and to access. The Internet also allows translators to reach and query subject matter experts, exchange information and views with colleagues, as well as to work virtually anywhere with ever more powerful and economical instruments.

This is why, today, the marginal utility of translation is even lower than twenty or thirty years ago.

Also, twenty-five years ago or so, success depended largely on the choice of a location for your business. Competition was larger and harder in a big city, but the customer base was larger as well as the availability of resources, starting with libraries. On the other hand, in a small town competition was weaker, but customers and jobs were fewer. New technologies and the easy and cheap access to an increasingly pervasive Internet chased from the market those who were unable to take advantage of them. Since then, the arrival of the first generation of digital natives has finally marked the end of the geographical element as a success factor, and translators have been further selected according to skills and technical resources.

Nevertheless, the industry and market structures have remained essentially unchanged for half a century or more, and customers address the market virtually in the same way, shifting from the Yellow Pages to Google.

In an interview given in 2010 to Phil Goddard, Henry Dotterer, founder of Proz, explained that the “market” side of ProZ is not different from a telephone book, in spite of the 224,000 jobs that have been posted since its foundation in 1999.

For example, in Italy, in 1982, the entry fee for a translator was roughly Lit 1,500 for a page of 60 strokes for 25 lines (approximately 250 words), requiring about 40 minutes to complete, while average salaries were around Lit 600,000 a month. Dictionaries were essential and often represented an expensive investment to make thoughtfully: they were few and poor, and a library was generally the only place for a research.

About ten years later, the fee of an experienced professional translator was Lit 20,000 for a page, while the average salary was about Lit 1,500,000 a month; the PC era was thriving and the most advanced translators could use a modem to send job files to customers and chat with geek colleagues via FidoNet, but floppy disk deliveries were still a plus.The important thing, though, was that, like ten years before, languages were still the heart of the service, or the service tout court.

Twenty years later, the average price per word for a professional translation is € 0,08, corresponding to about € 20,00 per page, with average salaries of around € 1,200 a month.

The Lit 1,500 per page of 1982 corresponded to a monthly salary of Lit 396,000, and roughly correspond to € 980,00 today.

Based on the same criteria, Lit 25,000 in 1992 was equivalent to Lit 5,280,000 of gross salary, which roughly corresponds to € 4,500 today.

Today, to invoice € 4,500 per month, a translator must produce at least 1,900 words a day. The tools available today make this possible, but they are a significant investment even for a professional translator, who also needs invest at least one third of his/her revenues on professional development classes for extra-linguistic skills.

In 1982, the bulk of spending was for paper, ink ribbon, and liquid paper.

In 1982, word rates were the only practical and consistent metric; today it is just simple and inexpensive, and allows for paying value-added services to the price of the now commoditized language services.

In fact, translation is commoditized and globalized, but this is no news in many other intellectual professions and industries, even professions younger than translation and the translation industry. An inversion of trend can hardly be reasonably expected.

Before waving the standard of a new period of de-massification of translation, with the renaissance of the artisan translator, and the surge of quality over quantity, one should first define “quality” and ask whether any customer, possibly in the manufacturing industry, rather than in a boutique banking business, is willing to pay more for less. Then, some hints on the alleged recovery of the status of the profession would be welcome. But if translation should still be viewed as an art, then one would better resign to a bohemian life, since the percentage of successful and healthy artists enjoying fame and money is less than minimal.

By the way, when quoting Jerry Mc Guire, one should keep in mind that he lost his job and, despite the fictional happy ending, nothing indicates that things will be different from what have always been, and even less that they could be how he envisioned in his mission statement.

In What Money Can’t Buy: the Moral Limits of Markets, Michael Sandel recalls how the same Larry Summers above cited Michael Lewis’s Moneyball: The Art of Winning an Unfair Game as illustrative of an “important intellectual revolution.” In Lewis’s book, an underfinanced underdog team uses its wits and the tools of modern econometrics to compete with rich, powerhouse teams. According to Summers, the lesson was: “What’s true of baseball is actually true of a much wider range of human activity.“ In reality, Moneyball turned out to be not a strategy for underdogs, at least not in the long run, as Oakland Athletics last made the playoffs, but haven’t had a winning season then, not because moneyball failed, but because it spread. After all, in John Maynard Keynes’s words, “The long run is a misleading guide to current affairs. In the long run we are all dead.”

The actual Moneyball lesson is that when talent is priced efficiently, the teams with the most money to spend on player salaries can be expected to do better.