← Back to portfolio
Published on

Dexonline

Few facts to understand the story: Romania has a so-called “language authority”. The Romanian Academy handles everything related to language and is the only state institution that is allowed to publish the “DEX” (Dicţionarul explicativ al limbii române – The Romanian Explicative Dictionary, an official Romanian language guide with word definitions). It is also the only institution that can officially add or remove words from the vocabulary or establish that certain forms of certain words are correct or incorrect. The Romanian language is very strict and has multiple rules, as many other Latin languages. If a word is not in the DEX, it is considered unsuitable for literary use. Globalization filled the Romanian language with countless English words and word roots which the Academy is not keeping up with. Nowadays, teenagers and people who work in multinational companies tend to alter the Romanian language by mixing it with English words. Thus the language used by certain categories of people is becoming more of a Latin dialect of English. Linguists fear that in less than 50 years, the Romanian language will completely disappear because its only speakers, less than 25 million inhabitants of Romania and the Republic of Moldova, are abandoning it. The last edition of the DEX was published in 2009 and the Academy hasn’t released anything related to it since.

It’s the end of August 2001, and Cătălin Frâncu, a Massachusetts Institute of Technology (MIT) sophomore, is paying a short visit to his brother in San Francisco. He sits at his computer which he brought from Boston to the West Coast. He wears round glasses that reflect the white empty display while his thin fingers play gently on the keyboard. N-E-G-A-T-I-V-E. The cursor blinks. Cătălin leafs through the Romanian Explicative Dictionary until he gets to the letter N. Then recites the definition of the word he wants to type and adds it to the database he’s been working on for a month. He opens a purple webpage with an empty bar and searches the word he added. His search shows one result and, on the right side of the screen, the definition of the first word in the first online Romanian dictionary: “negativ (Romanian for negative) – adj., expressing denial or refuse, something that denies.”

* * *

Ten years have passed since Cătălin Frâncu created the database Romanians now call Dexonline and friends have stopped telling him that an online Romanian Explicative Dictionary (DEX) is just a dream. A decade later, he and people around him started asking themselves how far a Romanian online dictionary can expand while the DEX editing authority, the Romanian Academy, does nothing to contribute to the development of the project. Dexonline has now reached its maturity. It’s the highest ranked website in the Education category on Trafic.ro, the largest web traffic counter in Romania, and one of the daily instruments that Romanians use everywhere around the world. It has over one million users each week, 28 admins, and over 400,000 dictionary definitions.

Cătălin Frâncu (33 years old) is a software engineer and comes from a family with a strong IT background. His parents were among the first generation of employees at the Romanian Institute of IT Research and Development, where they created and tested software on computers that filled an entire room. As a child, Cătălin memorized pretty much any information that made him curious. He used to keep his mind busy with computer games, either by creating them or by adapting them so they would work on different types of computers. During high school, he never missed an Informatics Olympiad, for which he used to prepare all by himself. Three times he managed to advance to the national phase of the contest. In his senior year, he won the silver medal at the nationals and went on to write a book on the psychology of programming contests. He also tutored a group of students passionate about informatics who wanted to compete in international contests. He only dealt with the Romanian Language Olympiad briefly, during his early years in middle school.

When his brother left for the United States to begin a Ph.D. in computer science, Cătălin was a freshman at the School of Automatic Control and Computers in Bucharest. At the time, he was saying he would never leave his country, the place where he was born, regardless of what the future held for his brother across the Atlantic. He declared his intentions to his family and to his friends. He even stated them on a TV show. “They asked me if and when I was planning to relocate since my brother was doing well in the States. I said I would never do that, but, three years later, I was hanging Romania’s map in my room at MIT.” During the three years he spent in university in Bucharest, he failed five exams. His GPA was a bit higher than  mediocre and nothing he studied felt like it would help him in the future.

In the U.S., Romanian friends of Cătălin’s spoke something he used to call “RomEnglish”. This was a mixed language, a combination of English words read phonetically, the Romanian way, and given Romanian meanings. “A colleague of mine once said he is taking his “car-ul” (“car” – English for “automobile”; “-ul” – Romanian definite article for singular masculine nouns) to the mechanic. Another one said he bought “tool-uri” (“tool” – English for “instrument”; “-uri” – Romanian definite article for plural neutral nouns) for barbecue. I had a friend who did a great impression of them and looked for those expressions that sounded awful if you translated them word by word.” All of a sudden, people at MIT were fighting over words and their meanings and forms. The controversies led nowhere because neither side possessed the ultimate argument, which was the Romanian Explicative Dictionary. In the spring of 2001, Cătălin met with a Romanian colleague in the dorm kitchen for a cooking session and told his colleague he needed more “crătiţi” (a Romanian plural for “pan” considered incorrect) to finish everything. His colleague stared at him and told him in a squeaky voice: “Dude, there’s no such thing as crătiţi. It’s cratiţe.” That summer, when Cătălin visited his relatives in Bucharest and bought a DEX, he realized the guy whom he’d been cooking with had been right.

In 2001, Romania had around one million Internet users, and the majority of the web pages were static, meaning they didn’t use databases and their content was updated only by a programmer, not by external users, as it happens today with websites that have administrators. In his dorm room, Cătălin had created a web page that allowed him to explore the mechanisms of a dynamic interactive website, which every user could add words to. The online Merriam-Webster dictionary and the DEX he bought from Romania inspired him to create a word search engine, a database with definitions he uploaded himself.

The dictionary’s first online address was voronet.mit.edu/dex and its server was Cătălin’s computer from MIT. (He later changed it to dex.francu.com and to its current domain, dexonline.ro.) In the beginning, this was a simple page with just a bar. You could enter whatever word you were looking for and press the search button. You could also read a story of how the dictionary was born and the reasons it mattered. In the beginning of 2002, the database had only 42 words, out of which Cătălin had introduced 34. The rest had been added by friends, MIT colleagues, and people who’d found out that Frâncu had an online dictionary.

“There were no plans in the beginning. The dictionary was just a project whose only task was to grow. It just stood there online. I’d check it every morning to see if anyone had added words during the night. I sometimes added one or two myself if I had the time. My first goal was to introduce words that started with letters that Romanians don’t use a lot, those words that take only a page in the Romanian dictionary, starting with k, q, w,  or y. The first users didn’t find anything they were looking for. They were bored and left and didn’t come back. Mostly everyone was convinced that this was not a project that was going to catch on. Many people actually said that to me in various ways. I realized I could get this project to do something only when I got help from others who forced me to make the development of the dictionary my daily task.”

But when Cătălin started to take it seriously, his enthusiasm lowered. For the online dictionary to exist he needed the original DEX in a digitalized form and only the Romanian Academy possessed such a document. The MIT project was independent and had no connection with the institution that printed the DEX. Cătălin wrote the Academy many times telling them about his dictionary and plans of converting the paper DEX into an online database for everyone with internet access to use, but he only got unclear answers. The most straightforward was the one he got from the Research and Artificial Intelligence Institute. They told him about the existence of a similar project in the Academy. Cătălin sent them eight emails. He was never told to stop, nor that he would someday get the help he needed, which was the dictionary in a digitalized form. More than 60,000 definitions awaited to be added manually. By adding two to three words a day, the online conversion would have taken Cătălin around 50 years.

* * *

On a February evening in 2002, the president of the Romanian Scrabble Federation, Ştefan Pall, was letting the organization’s online community know that a Romanian explicative dictionary was available on the Internet, “a project that could be very interesting for Scrabble players too.” Seven people responded to the forum post, three of them saying that the project was silly. The four others believed they should investigate more and see who was behind the project and what their intentions were.

The Federation was eager to update the Official List of Words (LOC), used during games of Scrabble in order to prove the existence of certain words or word forms. The old list included the 1996 and 1998 editions of the DEX, adding up to a total of around 60,000 words, as well as a list of declensions and conjugations. But the Romanian language has more than 140,000 words, and those missing from the LOC were either variations that did not appear in the DEX or words that appeared in other unofficial dictionaries.

Radu Borza, a Mathematics and Electronics graduate from Timişoara (a major city in the West part of Romania), who owned more than 50 dictionaries and loved grammar, was among the people who responded to Pall’s message. Borza befriended many people at the Federation while he was doing a Ph.D. in Linguistics in Italy because he didn’t communicate that much with other foreign students. He usually stayed in and played Scrabble online. He was using the same platform that was being used by many members of the Federation, including the president. That’s where he found out that the Federation was planning to re-edit the list of words and offered to help the technical commission that was going to decide how the new document was going to look like. At first, Radu was reticent about the online dictionary saying that “there’s a much more advanced project in the Romanian Academy, a smart dictionary that performs searches,” while the MIT dictionary was just at the beginning and couldn’t keep its independence if it intended to develop. (The Academy’s project was never launched and is now available only for the institutions’ members.)

Radu needed such list of words himself, even though he wasn’t member in any professional club or federation. He dreamed about creating his own version of Scrabble because he enjoyed games. “I lived my childhood during the 80s, when logic games in Romania were the rage. Every month there was a new game on the market, new ones or replicas of foreign games. In the summer, my whole family would gather around a playing table in the yard. We played rummy in front of the house that Ceauşescu tore down. When the mosquitoes were getting too annoying, we’d take the whole table into a big stone hallway. We used to play bridge on the same table. My brother had taught me how to play. He also taught me Scrabble and he always won. We had a very nice Scrabble game. The letters would stick to the surface when you pressed them. We paid an enormous sum of money for it at that time, but I still don’t regret it”.

After many debates about the content of the new list of words, the Federation members concluded that the document should be more complex and that it would be more useful to create a dictionary containing not only definitions, but also every form of every word. Ideally, it would be available online, so that everyone would have access to it and the printing costs wouldn’t be an issue. For the 140,000 words in the list, there were about 1.4 million inflections.

Writing the code would have been the most time consuming, therefore more than 200 volunteers from the Federation, Scrabble and crosswords aficionados from Romania, Romanians in living at home or abroad, and people with different professions and interests contributed to Cătălin Frâncu’s dictionary. They helped it expand by adding all the definitions from the 1998 DEX. At the same time, Radu Borza, who got over his initial skepticism, put to use his collection of dictionaries and introduced definitions for the top searched words on Dexonline that weren’t listed in the Academy’s dictionary.

By the 1st of August, 2004, the Dexonline volunteers had finished collecting and uploading the 1998 edition of the DEX, amounting to 65,531 definitions, 1,200 pages and 13.5 million characters (three times more than what the Bible has). “The Dexonline project debuted three years ago governed by the thought that ‘it’s about time’,” Cătălin wrote on his blog at that time. “For every Romanian in Romania and for those outside the country that want to speak their language a little bit better, having a dictionary on the Internet is not a luxury, but a necessity. Someone had to make the effort and we were the ones to do it. Unfortunately we were forced to choose the less beaten path, which meant retyping every definition from scratch. We started the Dexonline project on August 28, 2001, having serious doubts that we’d be able to find enough people with the devotion and free time to contribute to this effort. The enthusiasm of the dozens of people outmatched any expectation that we had. Dexonline is not perfect but it was and is the answer to a serious demand that no one else has tried to satisfy. People ask for a dictionary with an active presence online, and although we are not the perfect people to provide them with that, we fulfilled their request. It would be most useful if somewhere on the Internet there would be a dictionary taken care of by linguistics experts, but until that time, we hope that Dexonline will serve its public decently.”

The following years, thanks to the impact of hundreds of volunteers, Dexonline received as gifts many more dictionaries in digital format (a synonym dictionary, an antonym dictionary, a neologism dictionary) and topped 200,000 definitions. In August 2006, they launched DexFlex, a new component that produced a complete list of Romanian words and their inflections. 2010 marked the birth of m.dexonline.ro, the mobile version of the dictionary, and that same year the design changed to its new colors. Dexonline also got Facebook and Twitter pages.

Out of the 200 volunteers, around 40 stayed committed to the project. All others added a word once every few months or worked on recommending the dictionary to their friends. Radu Borza (37 years old) is one the few loyal ones. He gave his input on every aspect, from coding to debates about dictionaries that should be imported. He worked side by side with Cătălin to adapt them to the Dexonline format. He made the structure for DexFlex and created display options for definitions with diacritics which usually pose displaying problems. Once named website moderator (he and Cătălin are the only ones with full rights), he became the main linguist of the Dexonline team because of his passion for etymology and word relations. For every definition that a user regards as ambiguous, Radu now looks for a more understandable one in his pile of dictionaries. He has the necessary patience to offer explanations about the origin of specific words whenever an angry user expresses his disapproval about an incorrect or discriminatory definition. He gives numerous examples and clarifications beginning with their origin, relations between their definitions and the context in which they usually appear. In the end, he might even tell a joke about those words so you’ll always remember them.

Today, Cătălin Frâncu thinks of Radu Borza as one of those people who gain your trust because of their immense passion and dedication. If he would ever decide to retire from Dexonline, he believes Radu would definitely be able to take over everything, from organizing to fully coordinating the project.

* * *

In 2009, Dexonline had almost 300,000 word definitions and over 700,000 users. Dorelian Bellu (36 years old), a journalist back then, used it whenever he wrote an article. He’d been doing so for more than four years when he ended up suing a media company that refused to hire him as a writer. He took a Romanian language test in order to get the job, but failed the exam without being told his grade. He believed he’d scored almost 10 out of 10 and to pass the exam, he needed a 7.

The judges asked the plaintiff for an expert to analyze the test and Dorelian called the director of the “Iorgu Iordan” Linguistics Institute to evaluate his paper. The director told him he didn’t want to get involved because every judge had enough grammar knowledge to deal with the problem and they didn’t need experts to explain every comma. Also, if he accepted to examine the paper, he’d face tens of other requests coming from lawyers who’d want the punctuation in every document checked beforehand.

Dorelian went home and printed hundreds of pages with definitions and grammar rules found on Dexonline that he considered useful for the case. He put all papers in a file and took it to court. When the judge asked him where his lawyer was and why he was alone, Dorelian handed him the file and told him bluntly that he didn’t need a lawyer: “Dexonline is my lawyer.”

One year later, Dorelian noticed there were no recent posts on Dexonline’s Facebook page and wrote an e-mail to Cătălin Frâncu asking him to pay more attention to the social media account. He also offered to become the administrator of the dictionary’s blog and take care of everything related to marketing and online communication. He’d read about Cătălin in newspapers and he’d spoken to him a year before that when he’d written him to tell him he wasn’t satisfied with the website’s search options. “He answered me the next day,” Dorelian says. “He’d fixed it. I realized then that these guys took Dexonline very seriously. Maybe he didn’t do much, maybe he just pushed a button or something, but the way he took care of the matter made me look at them differently.”

Dorelian now works in advertising and social media. He has a degree in naval constructions and another one in communications. He believes his ten years of working in the media have been very useful because he learned that in order to not make a fool of yourself you need language skills. He enjoys playing chess and talking about politics and he sometimes writes for his blog or for different magazines in Romania. In his spare time, he tries to put together a strategy to make Dexonline popular and get investors to donate money to make it grow. In the last 10 years, Dexonline has made money only because of web banners and journalists have rarely written about it. Neither Cătălin nor Radu have been too interested in making a step forward and participate in TV shows to tell Romanians more about Dexonline. “Radu and Cătălin want to promote themselves, but they are both engineers, IT guys. They want to do stuff, not brag about it,” Dorelian says. “They haven’t paid too much attention to their public image. And this is cool nowadays, but also unusual. While everyone wants exposure, they wanted to create something to last, an academic instrument that will always be there.”

They all need money now, because they want Dexonline to be more than just a dictionary full of definitions. They want to expand the project, involve linguists and software engineers, buy new dictionaries and add them all to a huge database that will incorporate not just the DEX, but also synonym dictionaries, idioms or every new word that Romanians use right now. Adding these new features will not be cheap. Cătălin, Radu and Dorelian want Dexonline to help both those who speak Romanian and those who wish to learn the current Romanian language without using a dictionary that’s been on the market for two years. The Academy printed the last DEX in 2009 and hasn’t released anything else related to it since then.

Dexonline has to be better than the paper DEX, they say, and identify the day-to-day changes in the language, because the internet moves faster than the print industry. It also has to include mistakes that people make when using Romanian (The DEX doesn’t include incorrect words and their proper explanation such as many important dictionaries in the world have), new words with proper guidance on how to use them, and many more definitions for the same word, gathered from many different dictionaries.

Overall, Dexonline has to be a virtual language encyclopedia. If tomorrow Romanians would start using a new word, the linguists that Dexonline would hire would have to identify it quickly as part of the spoken language and create a definition for it judging by the context in which it would be most frequently used. Aside from linguists, Dexonline needs people to work with complex programs that would be able to look for these new words online in the media or in books available on the Internet. There shouldn’t be a word in the Romanian language that isn’t on Dexonline. “This way, Dexonline will always be updated,” Cătălin says. “And it’s for all people’s sake. Even famous linguists and dictionary authors want to get closer to the project because Dexonline allows them to make public their work in just a couple of days, while editing another dictionary could take years.”

“We could even exaggerate and say Dexonline could be a platform that will save the Romanian language,” Dorelian says. “In a couple of decades, Romanian will no longer exist or it will suffer dramatic changes because we are pushing into the language many foreign words and expressions. And it doesn’t even have many speakers. According to linguistic theories, if the language disappears, this severely affects the culture to which that language belongs. There will be no way of expressing particular feelings or customs that belong to the Romanian culture. Dexonline is the only one that can keep up with the language and record its changes. It is also the only tool that can give an example of what the Romanian language means and how it should be spoken, when the Academy’s paper DEX can no longer do that because it is published at a very low frequency. Therefore, in the future, Dexonline must tell you how cool it is to be Romanian and speak Romanian.”

* * *

It’s the end of September 2010, and Cătălin, the man who created Dexonline, gazes at his computer’s display, perched on a big red plastic ball in his cold basement from his Bucharest house. He types fast, then stops and gently taps the backspace key, while writing the Dexonline statement of purpose which he decided to share after almost ten years since he created the modest website at MIT. “Sometimes, aside from doing things, it’s good to explain why you do them,” he says. “We didn’t get here by taking the best path. It would be insane to say we did it all perfectly from the very first day. We could have gotten here faster, working less, having more money and more dictionaries. Surely there are people and institutions that could have done it better. But the reality of life shows us that he who builds a project is not the genius who studied a lifetime how to do it, nor is he the businessman who invests a lot of money in it. In the end, the project belongs to those who start working on it.”

Photo credit: George Enache/Esquire Romania