Introduction

Among LISA’s key benefits are its conferences and workshops offered through the Forums and Round Tables every few months in different parts of the world. They are LISA’s means of providing its members and other industry players with opportunities to learn and network (in good, old-fashioned face-to-face mode), and most importantly, to explore important issues via collective brainstorming. This sharing of ideas continues long after each LISA event as many attendees download detailed summaries and presentations, available immediately after each event.

Academics also leverage these conferences by presenting papers with the added intent of increasing their number of publications – an important measure in university circles. Considering that the ROI (Return on Investment) is still a magic word in the business world, the GILT industry included, I’ve no doubt that conference-goers are scrutinizing ever more carefully the exact value of attending each conference. The shortage of funding is being experienced not only in academia, but also by industry management, who often view conferences as non-essential activities. In this climate, the art of determining the value of conferences is increasingly important.

During the last two weeks of May, I participated in three conferences – two in Dublin, Ireland and one in Halifax, Canada. Because of their proximity in both timing and subject matter, it was ideal to compare and contrast their content while asking myself whether or not they were a good investment (please refer to the table below). This article addresses the value of attending conferences as a sub-theme while focusing on some major issues for the GILT industry:

1. controlled translation (EAMT/CLAW)
2. theory vs. practice (IJET)
3. globalization vs. localization; new vs conventional translation (CATS)

Table: EAMT/CLAW, IJET and CATS at a Glance

EAMT/CLAW
A New Beginning for an Old Concept: Controlled Translation

Background

The European Association for Machine Translation (EAMT) hosts various MT related events along with its parallel organizations, i.e. the Association for Machine Translation in the Americas (AMTA) and the Asia-Pacific Association for Machine Translation (AAMT). The EAMT conference has been an annual event while Controlled Language Applications Workshop (CLAW) was initiated in 1996, offering bi-annual series of workshops.

Some readers may recall that the Globalization Insider’s European Editor, Andrew Joscelyne, in his commentary at the beginning of the year, flagged the EAMT/CLAW conference in Dublin as one of the most important of the year. The conference was unique in combining the 8th EAMT gathering and the 4th CLAW to provide more explicit links to MT and Controlled Language. Indeed, controlled authoring and consideration of translatability of the source content are capturing increasing interest from the localization industry, and particularly from high volume translation/localization users.

The conference gathered some 100 experts on Controlled Language (CL) and MT from industry and academia with the former making up 57% of participants, indicating strong industry interest in these topics. In terms of geographical distribution of the participants, there was a strong domination of European delegates (56%) outside of Ireland. Of the 25 papers presented, only four speakers were based outside of Europe (in the U.S., Canada and Australia). The lack of representation from Asia both as speakers and participants was quite conspicuous - a likely influence of SARS and the post-Iraqi war situation [1].

Assuming that not all readers are familiar with CL, I would like to summarize a few key points about CL:

1. Controlled languages are subsets of natural languages, but their lexicon and grammars are restricted so as to reduce or eliminate ambiguity and complexity.
2. CL can be divided into two types:
   1. human-oriented CL (HOCL), which is designed to improve readability and comprehensibility of text for humans (particularly non-native speakers)
   2. machine-oriented CL (MOCL), designed, for example, to improve the translatability of text by computer, i.e. MT.
3. The earliest application of CL goes back to the late 1970’s with the Systran implementation at Xerox Corporation. The CL concept also relates to domain-specific MT systems such as Canada’s TAUM Météo (operational since 1978) whose success is based on the limited vocabulary and phraseology of the input text.

Content

The opening keynote address was given by Lou Cremers of the Dutch firm Océ Technology. As manufacturer of a range of office equipment such as printers and plotters, the company is a volume user of translation services. This justifies an internal translation section, although it also outsources some of its translation requirements. It produces a total of 2.5 million words of English source text annually, which in turn is translated into 23 million target language words (as of 2002).

With such a huge volume, Océ Technology’s goal was to optimize the translation process by introducing translation technologies, including MT, TM, terminology management and CL. It chose an off-the-shelf, controlled English product (MAXit) and integrated it into the company’s sophisticated translation workflow system.

The benefits of introducing controlled authoring was highlighted in achieving

1. terminology consistency
2. grammatically correct and succinct source text (a 35% reduction in word count)
3. improved performance of MT, along with the use of XML and a content management system
4. a reduction of superfluous information.

The speaker demonstrated concrete evidence of the contribution made by the use of a controlled authoring environment. It was emphasized, however, that CL does not solve all authoring problems and that training is extremely important for it to be effective. Océ also learned that one cannot really judge the value of a CL product properly until it is actually implemented, making pre-purchase product comparisons difficult.

This set the scene for the conference with compelling evidence of the benefit of CL from an actual user. The ensuing sessions were organized into successive, single-stream sessions under the three main groupings of Controlled Language, Machine Translation and Controlled Translation. The following section outlines some of the papers that focused on CL.

Sharon O’Brien from Dublin City University (DCU) presented some of the findings from her doctoral work-in-progress for determining universals for CLs (including both HOCL and MOCL). She compared eight controlled English rule sets:

1. AECMA Simplified English
2. Attempto Controlled English
3. Alcatel’s COGRAM
4. IBM’s Easy English
5. GM’s CASL
6. Océ’s Controlled English
7. Sun Microsystems' Controlled English
8. Avaya’s Controlled English

O’Brien’s results showed that the only rule common among the eight was the use of short sentences. Her findings were subsequently referred to by various other speakers, indicating that the results were somewhat surprising.

Ursula Reuther of the Institute of Applied Information Sciences focused on the relationship between readability (by humans) and translatability (by machine) of text written with CL in a technical domain. The most important rule category for readability was found to be lexical (restrictions on general vocabulary). For translatability, rules advising writers to avoid ambiguous structures and ellipsis, and those governing stylistic issues were among the most important. Reuther’s findings indicated that the rules for readability and those for translatability are not too different, but the former are subsumed by the latter. In other words, translatability ensures readability, while the reverse is true only to a certain extent.

Teruko Mitamura from Carnegie Mellon University (CMU) described an interactive grammar diagnostic system incorporated in CMU’s CL checker KANTOO (CMU is well known for its knowledge-based MT called KANT, which uses a controlled language). The system provides feedback to the user (author), with a detailed diagnostic message to facilitate ease of conformance to given CL rules. This research was motivated by the speaker’s previous experience with authors not familiar with CL requiring detailed information when the system rejects input sentences. This also fit in with other speakers pointing out the importance of training for the successful introduction of CL.

Another example of an industry application of CL was by Airbus, as presented by Laurent Spaggiari. Airbus began a project in 1998 to develop a CL, specifically designed to improve the quality of warning texts to be read by pilots in the cockpit of Airbus aircraft. The project aimed at producing a writing guide for the text to improve comprehensibility. In the Airbus context, enhancement of safety was the essential driver behind the use of CL. The project team also had to take into account the fact that the warning texts in English would be read by pilots with different mother tongues.

Comprehension analyses were conducted with sample sentences among human readers from different groups (for example, airline, flight test and flight instructor pilots and populations outside the aviation domain), and statistical analyses were taken to observe general syntactic rules followed by people. A similar approach was taken with the comprehension of acronyms by pilots. The resultant CL that is precise and concise in terms of term selection, abbreviation, word order, etc. has proved to produce texts that facilitate quicker and better understanding.

There were a number of other interesting papers:

1. controlling MT generation in speech-to-speech systems where the MT output serves as a prompt for the user on each end
2. applications of CL in knowledge representation and reasoning
3. CL designed to disseminate Biblical materials in English to non-English speakers.

The closing keynote was given by Steven Krauwer of the University of Utrecht who is the Coordinator of EC-funded ELSNET (European Network of Excellence in Human Language Technology). He recalled the fact that there are many different approaches to CL (e.g., proactive and reactive) and suggested that the important parameters for CL in future would be:

1. the interrelationship with future MT developments
2. a better theoretical understanding of CL (both HOCL and MOCL)
3. the likelihood of developing multilingual CL generation based on symbolic input

Reflections

The conference performed a great service by shedding light on the under-reported topic of “controlled translation,” exploring MT and CL dynamics from the points of view of both researchers and actual industry users. It showed that there are diverse approaches to CL now being examined. Following the success of this conference, perhaps both AMTA and AAMT could try to incorporate slightly different angles for their future MT events. My wish is to see in these MT conferences a presence of the translator community that, as I sensed from the other two conferences covered in this report, is not keeping up with the latest developments, e.g., in the field of CL.

Because of the proximity of dates and locations, the EAMT/CLAW conference was also advertised on the IJET conference site, but I only met one delegate from the IJET also registered for the EAMT/CLAW (apart from Teruko Mitamura and myself). From him I heard a very interesting story about his client, a patent attorney in Japan. The attorney prepares two sets of patent text (in Japanese): one to be filed in Japan in Japanese and the other to be translated and then filed in English, thus written with translatability in mind. Apparently these two Japanese source texts are each written in a quite different manner. I found this discussion highly relevant to CL, given that the attorney is exercising a type of controlled authoring, albeit without the help of technology. A joint conference on MT, CL and Plain Old Translation may enable us to further probe the impact and role of technology on human translations (HT), and benefit both the MT and HT communities.

IJET
Practicality of Doing Japanese Translation in International Contexts

Background

The 14th International Japanese/English Translation (IJET) Conference was held at the Burlington Hotel in Dublin, Ireland. This is an annual event organized by the Japan Association of Translators (JAT) for Japanese and English translators and interpreters working in different parts of the world. JAT is an associate member of the International Federation of Translators (FIT) and has some 500 members worldwide.

This was IJET’s first time in Ireland. This conference normally attracts between 150 and over 200 delegates, but the number was considerably lower this year, particularly from Asia, for the same reasons as the EAMT/CLAW conference. A quick glance at the participants’ list indicated a wide geographical spread, including the U.K., the U.S., Japan, France, Switzerland, Denmark and Italy in addition to Ireland. Although both English and Japanese were official conference languages, most sessions were conducted in English. My taxi driver commented that it was the first time he had heard such a large group of Japanese speaking English well and Americans speaking Japanese fluently – he was intrigued to know what kind of gathering it was…

An interesting background to IJET is that its most regular participants also belong to the well-subscribed mailing list Honyaku that began life in 1994 for Japanese/English translators (JAT has its own members-only mailing list, too). So, in the early years of the IJET there were many first-time, face-to-face encounters among participants who had been exchanging numerous e-mail messages through Honyaku.

IJET alternates between Japan and an English-speaking country, thus slightly shifting the focus each time, depending on the locality. For example, this time there were a number of papers on Irish literature, Irish law and English-Irish translations of textbooks. Furthermore, according to the organizer, this year IJET made an explicit effort to invite non-translator, subject-specialists from a number of fields in which JAT members translate (e.g., patent, economics and international accounting) to provide background on issues being faced in each field. Based on the positive feedback from participants so far, the organizer is hoping to maintain this new practice for future IJET conferences. [2]

Content

My impression is that IJET’s focus is primarily to serve the interests of working translators and interpreters and academics to a lesser degree, so the conference program tended to be practically oriented (particularly the round table discussions). Still, it had a fair number of academically-oriented papers on technical and literary translation, as well as the history of translation, including

1. Translating Economic Texts with a Focus on Company Annual Reports
2. Story and Reality in James Joyce’s Ulysses
3. Interpretation and Translation in Nagasaki, 1641-1868
4. Discovering What Survives Translation True: Translating Seamus Heaney

A paper by Judy Wakabayashi of Kent State University on The Fictional Figure of the Translator surveyed how translators and interpreters are portrayed in fiction, and their conflict with reality (taking on the theme of “fictional turn of translators”).

Akira Mizuno of Rikkyo University, who is both a practitioner and an academic who set up Japan’s first academic association for interpreters, discussed a quality assessment of the rapidly increasing media interpreting in Japan.

The localization and technology related papers included The Role of the Translator in the Software Localization Industry Today by Brigid Corby from VistaTEC in Ireland. She covered extensive ground, explaining where the translator fits into the software localization process from an industry insider’s point of view.

A presentation by Yuko Miyata of Satellite Station, Inc. on technical know-how of computer-related translations covered tips for the use of translation tools. One of the issues covered was the importance of the use of style guide templates, which relates to the topic of Controlled Language.

One round table session was devoted to Translation Memory, presented by Doug Lawrence, Trados Country Manager for the U.K. and Ireland. While the session was intended to be generic rather than language-specific, the presenter was clearly thoroughly knowledgeable about the product. It proved to be a lively session, reflecting a clear interest in the technology, although it became evident that TM use for J/E is not as high as for other, i.e., European, language pairs. This also seems to relate to general delays in tailoring TMs to accommodate the Japanese language. It was refreshing to hear a tool company clearly acknowledge the limited scope of TM in terms of its usefulness, i.e., to certain types of repetitive texts.

The user-unfriendly aspects of previous Trados manuals were also pointed out – but apparently these are now much improved in recent releases. It was interesting to note that the biggest Trados customer is the EU with 2500 licenses. This just reflects the latter’s translation needs in terms of their suitability to technology applications in relation to text types and language pairs. It was confirmed that Trados works better from English into Japanese, than the other way around, due to segmentation challenges with Japanese text. One translator mentioned that Trados works far better than Déjà Vu when working with Microsoft Word files, while another claimed WordFast is superior to Trados for Japanese/English translations. As for networked-use, MultiTerm works on the Internet, but we were told that from 2004, Trados TM Server will be fully operational on the Internet to allow networked translators to share TM and to save valuable time. As a result of this session, I am convinced that more translators will start trying out TM.

In terms of technology-oriented papers, there was one paper on MT, Machine Translation: Development and Deployment Issues for Translators by Teruko Mitamura, a U.S.-based MT expert. Given the strong antagonism against MT demonstrated from time to time in the mailing list discussions on Honyaku, I was slightly concerned about the reaction to this session. However, in reality, the session not only attracted a sizeable audience, but those in attendance displayed an open-minded attitude towards MT. It is hard to tell how representative this may be in relation to the whole community of Japanese/English translators, but it was a good sign that translators are perhaps becoming more open to MT and interested in learning what technologists are doing.

Under the topic of technology, my paper, Into Hypertranslation: Abusive Translation or Something for the Future? was on the future of translation. It explored the implications of further developments of digital communications platforms, using a series of translation experiment conducted in computer-mediated communication and distributed virtual reality environments.

Reflections

IJET is an interesting conference, which differs from academic-focused translation studies conferences, or translation conferences organized by regional translation associations that tend to attract mainly local participants. Its international nature is also proof of how E/J translators are increasingly working in diverse locations (nearly half the JAT membership is now based outside of Japan). The membership factor and IJET’s link with the mailing lists seem to strongly support the idea of rotating the conference venues. There is a clear benefit for translators and interpreters in being able to mix with their counterparts based in different parts of the world because it results in their being exposed to a variety of issues considered important in different locations.

The new trend in involving subject-specialist speakers will add value to the important domain-knowledge requirements for technical translators. Contrary to the narrowness of the scope of the language pair involved (single pair as compared to plural European language pairs, for example), I found the nature of this conference multi-dimensional. Furthermore, because of its practical orientation, interest in technology applications (particularly tools) seems high. Apart from its obvious value to J/E translators, IJET is highly recommended to academics involved in Japanese translation as a place for a reality-check and for widening one’s horizons within the translation profession. Theory without practice is like coffee without flavor…

Readers must wait for the next installment to find out…

1. why Anthony Pym inspected the labels of his clothing,

and

2. why Bert Esselink flew straight to Weight Watchers in New York after the 16th Annual Conference of the Canadian Association for Translation Studies.

---

[1] I would like to thank my colleague, Dr. Dorothy Kenny of Dublin City University (DCU), for providing further information and reviewing my EAMT/CLAW report, for which she was one of the main organizers.

[2] I would like to thank the IJET organizer, Ms Stacey Jehlik for providing further information and commenting on my IJET report.