LEO's history

 The history of the dictionary

 

The following account provides a rough outline of the evolution of the dictionary. More details can be found in the Milestones for the years 1995 - 2006 and 2006 - 2011.

I The pioneers (1995 - 1999)

The idea to create a German-English dictionary, which was initially intended for internal use only but was subsequently developed with the new media environment and the World Wide Web in mind, was conceived by Achim Jung. It was inspired partly by a quest for new challenges in the field of IT and partly by a sense of "frustration" - existing German-English dictionaries did not yet contain entries for IT terms and finding the right translation was a tedious undertaking. Initially hosted on a machine of the Institut für Informatik at the TU-München and offering only basic content and features, the project soon gathered pace and saw its content and popularity rise. Already back then, leafing through dictionaries, which was tiresome and more often than not futile, held much less appeal than searching for a translation online with the help of new media tools, even if the latter sometimes also (still) proved to be unsuccessful.

While the dictionary was still in its maiden year, founder Achim Jung passed over work on the dictionary to Peter Kai Wimmer who received support in the years to follow from Mike Gendreyzig, Stephan Haslbeck, Marc Sihling and Christian Werner. During this first phase, improvements included a first attempt at a structured page layout, word lists were sent to volunteers for correction, forms were integrated for feedback and new entries, and a guest book was set up. Above all, however, what was initially a rather humble collection of words grew to a sizeable list of 170,000 entries. This shows that from the very beginning, the dictionary thrived on the symbiosis between the dictionary team and the dictionary's users who contributed many private word lists.

The end of this period was marked by the last of the pioneers leaving the field of research for a career outside the university setting.

II. Wechsel der Wörterbuchsoftware (2000 - 2006)

Along with the changes in the team came a change in server technology. Over the Christmas holidays 1999, Elmar Bartel developed the very first version of his dictQuery software which - after the expected glitches and teething problems - was put into operation in February 2000. The redesign of the software had become necessary because the constantly increasing number of queries had become too great for the initial infrastructure (based on Apache, Perl scripts and a Wais database) to handle. To give you an idea of the difference between now and then: while four queries per second used to be enough to bring the server to its knees, the current (2010) dictQuery version - after a host of modifications and revisions on its way to maturity - can now handle 700 queries per second with comparative ease.

The second phase was also a break from the previous content-management approach. Instead of voluntary helpers and IT experts with only limited language skills, the first qualified translator was hired to scrutinise the content with her professional eye (something direly needed after the previous years). This move was made possible by the TU München and the Deutsche Wissenschaftsnetz granting permission for the sale of advertising space on the LEO website to fund the position. Other milestones that were achieved in this period include the launch of the German-French dictionary, the vocabulary trainer and discussion forums.

III On the path to independence: LEO GmbH takes over (2006 - 2007)

Since 1 April 2006, the dictionary services have been provided by LEO GmbH. While the team and the LEO concept - of offering all content for free - remained unchanged, the whole setting of the venture changed on more than one level: offices relocated from the north to the south of Munich, entrepreneurial novices had to assume full responsibility for all 'benefits and liabilities' (with all the advantages and disadvantages that entailed), translators were offered permanent contracts to guarantee continuity in the expansion and maintenance of the dictionaries (difficult in the context of a technical university), competition had to be faced up to without any ifs and buts, and new server structures needed to be set up using a commercial hosting service. In short, numerous technical and entrepreneurial hurdles had to be overcome in the first two years until our lion felt settled in the new environment.

The major challenges on a programming level included:

• The integration of the German-Spanish dictionary.
• The complete redesign of the forum software which among other things required the change to a PostgreSQL database because the old software was no longer capable of handling the flood of postings.
• The expansion of the vocabulary trainer into a more powerful language trainer allowing users to set up study groups and upload their own exercises and lessons.

IV New dictionaries and improved server availability (2008 - 2009)

2008 saw the launch of the Italian and Chinese dictionaries after a long period of preparation. Chinese with its logographic writing system using either simplified or traditional characters (with no spaces between words) supplemented by Pinyin transliteration, was previously unknown territory and it took us two years to deal with all the problems the language presented. As we believe that dictionaries without enough content are of little use, we felt that one year's preparation for the Italian one was the absolute minimum. Unperturbed by all these challenging demands, the team was not afraid to take on the next project: Russian-German.

All the while, queries to the dictionary have continued to rise and have required us to pay increasing attention to the issue of server availability. We gladly accepted an offer of IP Exchange GmbH to set up a second server location in Nuremberg in addition to the existing one in Munich. While getting our services to run from two locations was no mean feat, being able to provide the service via two separate arteries and knowing that a complete outage is now impossible, was well worth the effort. A total of 14 servers currently handle the queries to the dictionaries, forums and trainers and at both locations, a load balancer ensures that queries are evenly distributed to the dictionary servers.

V Russian and new dictionary features (2010)

Even with a starting base as the foundation for our Russian-German dictionary, it still took us a whole year to edit the content and make the necessary technical adjustments. Apart from extracting the original content into a file format which we could work with, the issue of the Cyrillic alphabet, which is different to that of western European languages and consequently requires a different character encoding system, needed to be addressed, the original content to be tagged according to our notation system, the dictionary pages to be translated, and the forums and trainer to be set up .... This may serve as an explanation for why the 'wish list' of desired new language pairs grows much faster than our actual dictionary portfolio.

Another reason for the moderately-paced growth is the fact that much of our attention has been taken up by the new version of the language trainer which offers the possibility of searching for inflected words from within a text so that users can easily put together word lists for exercises they have uploaded. In the last one and a half years, we have been working hard on algorithms that can derive possible base forms from inflected words which are normally not included in the dictionaries. This, however, is a major project that is far from complete and will probably keep us busy for the foreseeable future. The new search algorithms enable us to offer users the option of searching comfortably for words from within a text. Using the dictionaries in the so-called text mode, a whole text can be entered in a text field and a click on one or several words in the field will bring up a list of translations. This means words do not necessarily have to be typed in the search field anymore.

Die Löwen der Wörterbücher (bis Mitte 2021)

Durch die Hinzunahme von immer mehr Sprachpaaren wurden die bisher verwendeten Löwen in den wohlverdienten Ruhestand verabschiedet - denn warum sollte Sherlock gerade für das Englisch-Deutsch-Wörterbuch stehen - und nicht etwa für das Englisch-Französisch-Wörterbuch?

Da wir auf Grund der herzlichen Komplimente (Ugliest damn Leo I ever saw) aber natürlich wissen, dass die Löwen bestimmt einigen Nutzer*innen fehlen werden, sind hier noch einmal alle in voller Pracht zu bewundern:

VI Apps - The future is mobile (2010/2011)

The first attempts at making the dictionaries not only virtual but also mobile were rather feeble - which is no surprise given the technical possibilities. Between November 1999 and April 2000 users could (if they were lucky) consult the dictionaries on certain mobile phones via WAP (Wireless Access Protocol). We ended the project on 15 April 2000, as documented in the "milestones" where you will find the laconic explanation: "Suspension of the WAP interface until upgrade of dictQuery to multi protocol functionality". And in fact we dropped the initiative without shedding a tear. In a second attempt in November 2003, we set up a short messaging service in cooperation with burdawireless GmbH. But expectations turned out to be significantly higher than demand and after many months in which we didn't even make enough money with the service to buy each LEO employee two scoops of ice cream, it was decided that we would drop it again - the phone charges, the cumbersome way of sending a query and then having to wait for the results as well as the data limit per SMS all proved to be insurmountable obstacles.

It was only when smart-phones came on the scene that there was any dramatic change: easy usability, the possibility of mobile Internet flatrates and larger displays also made solutions for mobile dictionaries appealing again. Since LEO's resources are generally limited, we needed to wait and see which operating systems would establish themselves in the long-term. By the middle of February 2010 we were able to offer the first mobile solutions for Android, Blackberry, iPhone and Palm. The run on the small downloadable applications (in short: apps) was in stark contrast to the disappointments of the first attempts mentioned above: in July 2011, we counted more than two million downloads for the LEO apps that were available at the time for the major operating systems.

Milestones 1995-2005

Milestones 2006-2011