In 1991 we finally started working with our first customer abroad, Novell UK, and since then we have been expanding our activity to Microsoft, Novell Corp., Oracle, to all our friends in Dublin, DLS, ITP and IDOC, and to many more. We have never let go completely of IBM translations and have also expanded into the Italian market with Sistemi Informativi's traditional customers. A short parenthesis on our parent company. Sistemi Informativi is a major Italian software house, owned at 70% by IBM, with branch offices all around the Country. It develops software for central and local government, national agencies, finance and manufacturing, and creates applications packages which are marketed nationally. This means, as we will see later, that we have good technical support available all the time.

Customers and tools have changed, and we are happy to say we have expanded our basic mission of localizing/translating into Italian. Our organization, of course, has grown. We can now rely on 40 translators who all work in- house. This has meant a major effort in office space, hardware and software acquisition, but it has made all the difference in security and quality assurance. We find that there are still too many advantages in keeping the teams working together under the same roof. Project management costs are lower, machine translations operate better on an internal network and QA processes do not generate unpleasant surprises. There are of course ways of working well remotely, but they require trained personnel that is hard to find.

The Translation Center has also nine Project Managers (two at Senior level), an Administration, Technical and Test Support, and QA. Sistemi Informativi has set up the support of an Advanced Tools R&D department that so far has been able to answer all possible questions and a lot more. With this kind of help, we have developed programs for our network (automatic backup procedures, mainframe interfaces, ISO 9000 procedures on our server and so on), and have been able to understand the really complex software we have localized in the past years. I feel that our real added value lies in this technical support.

Our years with IBM have provided us with a methodology that we have adapted to our changed requirements, especially regarding the QA process. At the LISA Forum in Amsterdam last month, I was happy to see that we have been moving along lines similar to the QA Metrics system that has been developed by the SIG. We normally use an internal program that, given a description of the material to be translated (number of pages, layout, difficulty, topic and so on), extracts a percentage of pages to be evaluated, and prints out a form on which errors are registered and divided by type and level. Even before ISO 9000, our procedure foresaw that we simply could not deliver material to a customer without a complete QA cycle and approval.

In our Translation Center, the responsibility for quality lies in the hands of Project Managers. Our PMs do not translate. They manage what comes in from the customers and make sure that translators are fully consistent; they also review the material during the translation process to make sure that it will pass the QA tests. QA is a small and expensive department that must only certify quality. It does not review and correct software or documentation, but only states the quality level of the material analyzed. If it is good, it can be delivered to the customer, otherwise it goes back to Project Management for full review or retranslation. Senior PMs are also responsible for financial planning and budgets and they make sure that projects go through the production cycle only once.

Naturally, the growth of our Translation Center has encountered many problems. First of all, finding the right personnel has been quite difficult. We have tried many solutions: young people right out of university or language schools, expert translators from our competitors, technical experts with little translation skills, highly skilled translators with no technical experience, and every combination you can think of the above. Percentage wise, most of our staff was trained by us: given the appropriate time and money, we do prefer training young people with good linguistic background from scratch. In the past four years, we have grown of about ten translators per year and it has been very hard to find them. Just to give you an example, last fall we put an ad in one of Italy's leading newspapers; we received 450 replies, 20 persons were admitted to a training class, 10 started working with us, we will keep seven of them. The investment in time and money seems sometimes just not proportioned to the results. We have also tried to make agreements with language schools, asking them to include classes on computer terminology, but since conflicts of interest arise, we have dropped that possibility.

This leads to the problem of our competition. If the localization/translation industry in Italy were sufficiently well established, there could be a good exchange of personnel between companies. Instead, most of the business is handled by small translation agencies with little or no hired personnel. Besides the business with IBM (based in Naples) or Olivetti (based in Ivrea, near Turin), in the Country there is no steady flow of IT translations, and most translation agencies have very few, if any, free lances with computer related experience. Furthermore, most Italian companies still do not consider localization/translation as a cost they want to see in their budgets. The new EU regulations on deliverables with documentation will, I hope, one day make a difference. For the time being, projects are being given to very small companies that compete with each other by dumping prices which seem to have gone way below survival level.

When speaking to our Italian customers, it is very hard to make them accept the idea of an industrial process with high quality guaranteed, machine translations with memories that can be recycled and so on. They are all very interested, but when their Purchasing departments see that the job can somehow be done for half of the price, no quality consideration will hold. Eventually, when major Italian projects with high amounts of material to be recycled do come our way, it then becomes important to keep our teams sufficiently flexible to meet the requirements of the Italian and international markets. For example, last year we worked with the Italian Ministry of Defense and stepped into an entirely new world. Ten thousand pages of military equipment terminology and six months of work in a military environment make it difficult for any translator to quickly adapt again to software data bases or wordprocessors!

Furthermore, Italian companies and multinational companies have different stylistic requirements: the former want a text totally rewritten in good Italian, the latter a more literal translation that doesn't lead to any misinterpretation. The same translator finds it therefore rather hard to render text, either in accordance with the requirements of Italian companies implying good, clear language without repetitions, thus increasing the amount of words of a good 15%, or in his/hers next project for multinational companies go back to keeping text as literal as possible in the short 'subject, verb, object' style.

Whatever the style, customers want different translations for many basic terms: IBM and Microsoft are typical examples, and there are also other customers that have their requirements, thus adding on to the confusion. This is now changing thanks to a dictionary produced by the joint efforts of Microsoft Press and its Italian publisher, Mondadori Informatica. We have decided to use this dictionary as the basis for our terminology, and this has greatly helped us in various contentions.

We are now considering (and are very open to suggestions) a procedure that will help us group terminology and glossaries into a logical data base and are evaluating what parts of the material gathered over the years is specific to only one customer - and thus of no general use - or what is extendible to others. A good re- handling of our on-line dictionaries, indicating the preferences of our major customers at least for some basic terms, is an item that I would like to in this year's budget, and that I am sure I would recover in quality and consistency.

I hope that those who have taken innovative steps on any of the topics discussed will get back to me because I would enjoy very much having their opinions.