In his presentation to the LISA Forum – Europe held in Mainz, Germany in Spring 1997, Tom Lueck, CEO of Logos, Inc. identified the availability of market studies as one of the three acid tests of whether a translation market really existed or not (the other two were IPOs and external investment). By this standard, the language industry is still ramping up: although a number of surveys have been published in the past, their scope, quality, availability, absolute price and price/performance ratio have often left a lot to be desired, despite the in some cases considerable sums invested by national and supranational funding bodies, industry associations and individual organizations and researchers.

One problem is that in some cases, such as in many surveys sponsored by the European Commission as part of research projects, it is the researchers themselves who have done the surveying, although they may not have the marketing and/or business training or outlook necessary to produce workable surveys or meaningful results. However, there is now a growing awareness of such problems and of ways to solve them, as the recent SLIG/LRC workshop and the EAGLES workshop on tools evaluation held last November in Brussels go to show. At both events, relevant industry players in each case debated shortcomings in methodologies and surveys and means of improving the situation.

Another problem that affects language technology in particular is the relative immaturity of the systems and the lack of conventional commercial and industrial structures (many technologies and systems are only just emerging from academic labs). A corollary of this non-commercial approach is that agreed industry standards and methodologies for testing systems are only now being finalized in some areas, and are still outstanding in others. Such standards and methodologies in turn facilitate high-quality studies by providing commonly agreed baselines and reference checklists, among other things, and hence strengthening claims to objectivity. Conversely, actual and perceived methodological holes and shortcomings have damaged peer and market acceptance of many of the technical surveys produced so far.

Nevertheless, the success of open competitions such as that held by NIST (the US National Institute of Standards and Technology) on information retrieval, among other things, shows that it is possible to conduct rigorous testing and produce meaningful benchmarking data. This applies even across the academic-commercial divide and in fast-moving areas of technology. In Europe, much pioneering work on a general framework for language technology and resource evaluation has been done by projects such as EAGLES (European Advisory Groups on Language Engineering Standards), DISC (spoken language dialogue systems), SAM (speech assessment methodology), the EAMT (European Association for Machine Translation), and organizations such as ELSNET (European Network in Language and Speech) and ELRA (European Language Resources Association). Supplementing these are the more market- and application-oriented studies such as the Ovum Group’s LE 2000 (1993) and Natural Language Markets: Commercial Strategies (1991).

However, with both the tools and the services industry currently in transition to a more mature paradigm, there is a pressing need for more data, and for more market data in particular. How big is the market, what are the trends, and who is the competition? What requirements do and will customers have, and what approaches and solutions are available? With many millions of dollars hanging on these questions, it is unsurprising that the call for reliable market data is increasing. This is especially true given the chronic fragmentation and small average unit size of the industry: as the introduction to the Allied Business Intelligence survey, Language Translation: World Market Overview, Current Developments and Competitive Assessment puts it, “We have never before seen an industry so large but yet so disparate”. There is not even agreement on the overall size of the market, either internationally or for individual countries or sectors.

There are a number of reasons for this lack of a baseline in addition to potential methodological problems. The first is the nature of the market and the way it is perceived. Since much translation is performed by non-professional translators (secretaries, in-house specialists, copywriters, journalists, etc.) and is often either not billed as such or not billed at all, total volumes are generally impossible to define even within individual organizations, let alone at a macroeconomic level. In-house translation by non-translators, for example, will generally not show up on job sheets. In addition, in many cases, the lack of agreement on terminology makes apples-to-apples comparisons hopeless (see the article on this subject by Prof. Sue Ellen Wright elsewhere in this issue). Thirdly, the small average unit size of players makes any collation and evaluation of occupational records and the like a laborious process.

In this situation, the appearance or announcement of a number of new surveys providing up-to-date information on various aspects of the language industry is to be welcomed. In February 1998 the 1997 version of the LISA Industry Survey was released, and was followed in May by the Allied Business Intelligence production and the recent announcement of the Equipe “Globalization ‘98” studies. The three studies are briefly profiled below.

LISA Industry Survey

Title	The LISA Industry Survey
Publisher	LISA
Authors	1995: Ovum (Rose Lockwood) and LISA Administration 1996: Rebecca Ray 1997: Deborah Fry
Date	February, 1998
Format	Hardcopy and electronic form
No. pages	1996: 44 pages, 1997: 72 pages
Focus	Annual snapshot of issues affecting LISA membership
Price	Prices range between USD 50 and USD 450, depending on whether the hardcopy alone or the hardcopy and electronic version is ordered, on LISA membership and on whether a contribution was made to the Survey.
Strengths	Annual survey offers baseline for monitoring over time Detailed questions produce data on specific topics Electronic format in addition to hardcopy for easy dissemination and incorporation of results Customer, vendor and tools coverage (in theory) Reasonable price
Weaknesses	LISA world only Small, changing respondent population Questionnaire approach; little follow-up possible Uneven level of responses (ease of use problems/some confidentiality concerns) Questionnaire needs adapting to changing market Photocopied presentation format (hardcopy)
Overall impression	Mood check for the LISA community
Contact address:	LISA Administration, 7 route du Monastère CH-1173 Féchy Switzerland Tel: +41-21-821-3210 Fax: +41-21-821-3219 E-mail: LISA@lisa.org Web: www.lisa.org

The origins of the LISA Industry Survey date back to 1993, since when it has been conducted annually in some shape or form. Targeted towards the LISA constituency, the written survey as performed since 1995 is broken down into three groups –publishers (i.e. software and hardware manufacturers), service vendors and tools vendors. The questions focus in particular on workflows and structures within the localization industry, as well as concentrating on practical issues such as tool usage and quality systems. In addition, a number of questions designed to elicit information relevant to the industry as a whole (e.g. organization size, activities and geographical location, new business opportunities) are asked of all three groups. In a number of cases, the report also includes the results of other surveys conducted under the auspices of the Association. In the period since the Survey was first published, an attempt has been made to adapt the questionnaire format to the growth and changing needs of the target audience while still providing a common baseline on certain key issues to allow exploration of industry trends over time. However, the questionnaire and survey design could certainly be optimized further to reflect trends such as globalization, and to address the various different subgroups more precisely.

Data received by LISA is anonymized before the questionnaires are sent for evaluation, to ensure that confidentiality is preserved; nevertheless, this would seem still to be an issue for some potential respondents. In addition, of course, the sample population is both relatively small and can differ significantly from year to year. This means that the Survey should perhaps be seen as providing a “snapshot” or mood check for the industry rather than empirically hard and fast data. On the positive side, the data in the survey reports is specifically intended for dissemination and an electronic version of the results is provided to facilitate it. In addition, the price has been kept to a reasonable level (especially for contributing LISA members).

Allied Business Intelligence, Inc. Language Translation

Title	Language Translation: World Market Overview, Current Developments and Competitive Assessment
Publisher	Allied Business Intelligence, Inc.
Author	Marc Liggio, Associate Analyst
Date	May, 1998
Format	Hardcopy
No. pages	318
Focus	General translation market (bias towards tools and service vendors)
Price	USD 1,995.00 plus postage + packing
Strengths	Topicality Customer, vendor and tools coverage Handy (basic) directory
Weaknesses	Methodology unclear Sources not always given Way at which data arrived at not clear Quality of analysis Figures for total industry volumes seem low Confusing structure Polarization of human versus machine translation Uneven geographical focus Poorly written and edited
Overall impression	Collection of industry data plus comments
Contact address	Allied Business Intelligence, Inc. PO Box 452, 202 Townsend Square Oyster Bay, NY 11771, USA Tel.: +1-516-624-3113 Fax: +1-516-624-3115 Web: http://www.alliedworld.com/

The “new kids on the block” to the language survey scene are Allied Business Intelligence (ABI), an Oyster Bay, NY-based high-tech consulting company. Their study of language translation, hot off the press, is designed to offer a “world market overview and competitive assessment”.

Unfortunately, although the study offers a topical directory of and collection of data on key industry players, it falls short of its aim, and for a number of reasons. Firstly, the structure is somewhat confusing, making a strict distinction between human and machine translation (in practice, a much wider range of different but complementary technologies is used by both professional and occasional translators), together with a large and relatively unintegrated chunk of information on fax workflows.

Secondly, there is too little indication in the study itself of the methodology that ABI used or, in many cases, of the source of the opinions or figures quoted and the drivers on which they are based. Given the current uncertainty about even basic data such as the total size of the market, this was an opportunity lost.

To do ABI justice, additional information provided to LISA and made available to the author described the methodology used as “non-disguised judgment sampling, which really means that an analyst interviews industry players for preliminary information and juxtaposes and cross-references the results with secondary information from a wide variety of sources. Just like any military or government intelligence operation, our intelligence is compiled by triangulating information from multiple sources and having an intelligent and capable analyst determine the ‘truth’…We looked at a lot of other data on the subject and rejected quite a bit of it…We contacted nearly all of the major player (sic) and many of the minor ones, not to mention associations and other government agencies who publish data that helped us draw conclusions.”

Nevertheless, in a number of cases, a more differentiated analysis would have been welcome. Thus, for example, little distinction is made between the different types of translation company (the terms “company”, “agency” and “bureau” are used interchangeably). Equally, the effect of quality systems and process management – an area in which LISA has played a considerable role - is not sufficiently addressed. In addition, the “Strengths and Weaknesses” section of the company reviews is relatively heavy on the first category at the expense of the second.

The “Associations” entry on LISA describes its main purpose as “cooperating with industry partners to develop software localization production standards”, omitting any mention of the Forums as well as three quarters of the membership list. Equally, the list of German language associations contains four entries, two of which are in fact the same organization (the BDÜ) while another one - DIN, the German national standards institute - is only of subordinate interest. In fact, it is common knowledge that a number of other translation and terminology organizations operate in Germany (at least four of which are well established and one of which, ADÜ, has just received candidate FIT status).

Equipe Global Forecast/Globalisation ’98

Equipe Consortium Ltd.
31 Panton St.
Cambridge, CB2 1HL, UK
Tel.: +44-1223-350-340
Fax: +44-1223-722-044
E-mail: rose.lockwood@equipe.co.uk
Web: http://www.equipe.net/

Title	Globalisation ‘98
Publisher	Equipe Consortium Ltd.
Authors	Rose Lockwood
Date	August 1998 (first iteration)
Format	Hardcopy
No. pages	“under 100”
Focus	Sector-specific market data
Price	GBP 299
Strengths†	Sector-specific analysis Top-down and bottom-up approach Reasonably priced
Weaknesses†	No accounting for changing language flows within short-term forecasts Choice of priority segments open to argument
Overall impression†	An interesting study offering a high level of granularity
Contact address:
† No conclusive judgment is possible since the study has not been published and was not available to the author.

Produced by the doyenne of language industry surveys, Rose Lockwood, and her co-author Paul Crawford, the soon to be published Globalisation ’98 offers a more general approach based on what its authors call a “full-scale market model for the globalization and language industries”. According to Rose Lockwood’s talk at the recent SLIG meeting in Dublin, the study concentrates in the first instance on what it has identified as the premier global industry sectors (aerospace, automotive, business services, content publishing, engineering equipment, IT equipment, pharmaceuticals, software publishing, telecoms equipment and transnational organizations), the study is designed to update and replace the 1995 Ovum LE 2000 study by Rose Lockwood and David Lewin. It emphasizes market segmentation for studying language service requirements, in response to the extension of localization to different sectors, mirroring the current trend to vertical segmentation as well as globalization.

The survey unites a top-down macroeconomic approach using data from OECD, Eurostat, EIU, etc., with bottom-up interviews and desk research to produce “typicality factors” for trends such as language flows, volumes and outsourcing. Combined with production data, this was then used to calculate spending on documentation, localization and translation as a proportion of output per sector, country and export destination. The result is a series of tables illustrating the gross value added for key producer countries in each sector and percentage breakdowns for languages within sectors and countries.

While minor quibbles with regard to the methodology would seem justified (in par-ticular the forecasts for individual languages are static across the forecast periods (which are, admittedly, short), meaning that any changes in language demand can not be plotted), the survey would seem to offer a wealth of new and highly specific data. The authors plan to issue regular market updates and detailed sector studies, both individual purchases and an update subscription service are foreseen.

However, even if some of the spadework has now been done by these and other surveys, much still remains to do. Discussions within LISA in the context of the LISA 2000 initiative and at the SLIG/LRC Workshop have highlighted a number of areas in which further work is needed. One key area is salary data, since the SLIG/LRC workshop highlighted once again the problem of retaining key staff.

One problem here is that such data is qualitative rather than quantitative, and so needs to be gathered via expensive personal interviews and research. Even where the big six and similar consultancies are not used, USD 250-500k, depending on the qualitative focus, is required simply to start the ball rolling. Experience has shown that sponsors are hard to come by, even where a number of organizations share the cost. Care must be taken if we are not to remain in a vicious circle, in which data has to be sold at high prices in order to make ends meet, thus effectively excluding part of the market, and prolonging the “do it yourself” techniques of the past. It is time for everyone to join forces, agree on an agenda and prioritize desired results. Anyone interested in helping should contact LISA Administration, since the LISA 2000 Initiative has explicitly adopted this action item, and will be actively pursuing options both internally and in conjunction with other bodies, including hopefully SLIG/LRC.

---

Deborah Fry
Fry & Bonthrone Partnerschaft
Rochusplatz 10
55252 Mainz-Kastel, Germany
Tel +49 6134 22504
Fax +49 6134 22860
E-mail 100637.711@compuserve.com