6

INTERNATIONAL DECISION SUPPORT SYSTEMS

Many executives are choosing to internationalize operations to avail the corporation of larger and more fruitful markets, competition among labor forces, and economical location and distribution incentives. With internationalization comes geographical dispersion, increased industrial and market competition, and increased access to labor pools and natural resources. However, it also brings variations in the technical, legal, economic, and cultural forces affecting the operations and decision making of the enterprise, the impact of which is affected by the form of internationalization.

Transnational corporations can take on a variety of forms. For example, it is possible that offices in the various countries produce different products and are essentially separate. On the other hand, it is possible that the products are manufactured or created in one country and marketed in another. Or there can be some combination of the two, such as what Dyment (1987, p. 22) described:

The global corporation may have a product that was designed in a European country, with components manufactured in Taiwan and Korea. It may be assembled in Canada and sold as a standard model in Brazil, and as a model fully loaded with options, in the United States. Transfer pricing of the components and assembled product may be determined with an eye to minimizing tax legality. Freight and insurance may be contracted for relet through a Swiss subsidiary, which earns a profit subject only to cantonal taxes. The principal financing may be provided from the Eurodollar market based in London. Add the complexities of having the transactions in different countries, with foreign exchange hedges contract gains and losses that sometimes offset trading losses or gains, and one has a marvelously complex management control problem.

Decision Support Systems for Business Intelligence by Vicki L. Sauter Copyright © 2010 John Wiley & Sons, Inc.

279

280 INTERNATIONAL DECISION SUPPORT SYSTEMS

Another form of internationalization is described by Sankar and Prabhakar (1992, p. 251). This example involves not the production process but rather the sharing of data.

Consider the development of a Decision Support System that could support stock trans- actions for transnational brokerages with offices in New York, Rome and Frankfurt. Such a DSS must monitor the activity on multiple exchanges and in multiple markets to help the analyst determine what stocks to trade, when to trade them, and how to trade them. If the stock broker in New York wants to initiate a particular stock transaction, and if that company is listed on multiple exchanges, he or she needs to decide trading on which ex- change is most profitable. If for example, the decision is made to trade on the Rome Stock Exchange, the transaction is sent to a front end processor (FEP) in New York, which then transmits it to Rome using a private line. The Rome office sends a confirmation message to New York and sends a duplicate copy of the transaction to the head office. Further, the database used by brokers at all offices needs to be updated immediately so that models tracking trades and prices will be accurate. Clearly the coordination among these systems, while still providing decision support, is challenging.

Decision support systems have the potential for great assistance for multinational de- cision making because technical variability, legal innuendos, cultural differences, and eco- nomic pressures and their coordination exacerbate the turmoil associated with the poorly defined choice processes generally supported by DSS.1 However, if not implemented prop- erly, DSS can add to the problems of transnational decision making. In order to exploit the benefits, designers need to be sensitive to a wider variety of issues and problems than those considered in the design of domestic systems.

For example, there is reason to believe that there would be differences in preferences for user interface options for transnational systems. Understanding the preferences and their implications is crucial. Since the user interface is the only way one can interact with the computer, its acceptance by users limits the usefulness of the system as a whole.

The user interface can communicate the importance of information and modeling within a system. Different colors, size of representation (and relative size of representation), spatiality, and contrast provide the "nonverbal cues" for the user interface. Even the way in which one moves from screen to screen or accesses information carries some significance. That is, the user interface can convey what is important to the organization, how the "power" in the organization is controlled, or the corporate norms and expectations.

Consider the screen shown in Figure 6.1. In this screen, the financial implication of a proposed transnational corporate change to the United States is emphasized. The message is carried in two ways. First, the implications for the United States are the only ones that default as open to the screen. Users of the system are, in a sense, forced to at least see them (if not use them). However, the implication is that information regarding all other countries is "optional" to the decision because the user needs to take explicit action to cause those results to appear on the screen. The second way in which the United States is emphasized is through the size of the windows. Even after one has opened the windows for other countries, they are considerably smaller than the window containing the U.S.

1A team at the University of California at Irvine's Center for Research on Information Technology and Organizations studied the role of information technology in the economies of 11 Asia-Pacific nations. In countries where the investment in information technology exceeded other investments, such as plants and equipment, productivity was the highest. "This means IT investment is more productive than other investments," says one researcher.

INTERNATIONAL DECISION SUPPORT SYSTEMS 281

Figure 6.1. User interface implications.

financial data, hence conveying that the non-U.S. data are less important. A similar effect could have been obtained by displaying financial data only in U.S. dollars and not in local currencies. The implication of these differences is, of course, only a problem if the message they convey is unintended.

A second problem also is illustrated in Figure 6.1. In this case, instead of emphasizing a specific country, the size and default open options suggest the relative importance of particular analyses. As in the previous example, this screen design suggests that financial implications are the most crucial, whereas all other analyses are clearly secondary. This suggestion of the importance of particular steps in a typical analysis is also conveyed in Figure 5.31. In that screen, the system provides explicit encouragement for the user to attempt to change values and rerun the simulation. The availability of the option is making a statement about the importance of sensitivity analyses; the subtle recommendation would not be apparent without those automatic rerun buttons. This apparent support for particular options can present a problem for a transnational DSS when there are clear cultural differences in the modeling preferences across the cultures. Such differences will be discussed in later sections of this chapter.

Better user interfaces would have given non-U.S. countries greater representation on the screen. Perhaps no analyses would be open as a default, but rather the world as a whole is shown, and users can click on the country—or countries—of interest. Similarly, it would send less of a message if users needed to actually request all options.

The relative sizing and location of objects on the screen are not the only aspects needing attention in a transnational DSS. Since the user interface may be the basis for interaction with other managers using the system, users become totally dependent on this interface for prompts that would otherwise come from "nonverbal cues" and other tempering cues in communication. Hence, words lose their intonation and the user becomes totally dependent upon symbols and icons to convey more information. These new ways of affecting patterns

282 INTERNATIONAL DECISION SUPPORT SYSTEMS

of communication are fine as long as everyone agrees to the meaning of the various cues. Problems occur, however, if there is a difference between the "codes" meant by the creators of the cues and the codes used by the consumers of the cues.

In addition, the user interface may have a variety of problems associated with the use of multiple languages. Many cultures, such as the French, are adamant about maintain- ing their language as an active part of their culture, not just some quaint aspect of the small towns in the country. Hence, if one of the nations involved with the system is a country such as France,2 providing a single-language transnational DSS may be impossi- ble; translation of files, commands, databases, and so on, may be necessary. Translations can be tricky. Not only do the words need to be translated, but also the meaning of the words as a whole. For example, the Japanese interpret the word "pragmatic" to mean "tool user." Clearly, the meaning conveyed by referring to someone as "pragmatic" and that associated with "tool user" are quite different. Without an understanding of the lan- guage and the culture, the meaning of information used for decision making might be lost. As a result, translations can be time consuming and people consuming. While there are automated translators, they cannot be relied upon in such an unstructured setting; they rarely reflect the nuances associated with data. For example, consider the computer- generated translation shown in the box. Even without having the original Italian version, it is clear that the meaning of the communication has been lost through the translation of the words.3

2Even a system shared with Canada, a country quite similar to the United States, might require a DSS to employ multiple languages, depending upon its application. Since the French-speaking population in Canada is so numerous (especially in the Quebec province), Canadian law requires the use of both English and French in many circumstances. For example, even candy wrappers in Canada must provide all information, including the ingredients and nutritional information, in both English and French. 3Much work on language translation is in progress and some is much better than others. Even with the best of the software, though, one risks losing nuances in the meaning of words.

Design Insights The Toubon Law

In France the use of French is required by law in commercial and workplace communications. In 2006, GE Healthcare, a French subsidiary of a U.S. company, was fined €500,000 plus an ongoing fine of €20,000 per day for providing software and related technical documentation lo its employees in the English language only. The Toubon Law (the full name of which is Law 94-665 of August 4, 1994, relating to usage of the French language) requires French to be used in official government publications, in all advertisements, in all workplaces, in commercial contracts, in some other commercial communication contexts, in all government-financed schools, and some other contexts, including broadcasted programs. The Civil Court of Versailles followed a strict interpretation of the Labor Code and on January 11, 2005, ordered GE Healthcare to immediately provide its employees with (i) a French translation of its software and (ii) a French translation of documents relating to employee training, safety, and health instructions and training manuals. In addition, the court ordered the company to have documents relating to products already on the market translated into French by June 1, 2005, with a daily penalty for noncompliance of €20,000 per document.

The Toubon Law also allows for the fine of individuals caught adulterating the French language with commercial or official English, including computer terms.

INTERNATIONAL DECISION SUPPORT SYSTEMS 283

Even when the text is translated properly, its meaningfulness can be affected by the technology associated with data transmission if the language requires special characters. Often, if messages are not sent using an appropriate gateway, encodings become damaged or changed, and hence the message becomes garbled. Some transnetwork software strips off control characters, making the reading of text impossible. So, for example, rather than receiving Japanese characters, one might simply receive the following on the screen:

$NJ8>0$NNC$G!"$=$l$OEnglish$B$NJ8>0$91#J

The following was posted on an electronic discussion group dedicated to communication regarding historical issues, H-NET. It is included here to help the reader understand the problems associated with translation for transnational DSS.

Note from H-NET: Professor Andreucci, the moderator of H-ITALY, is fluent in Italian and English. H-NET asked him to review one of the new automatic language translation programs. His review appeared in Italian on H-ITALY. What follows is the automatic machine translation into English of his review. It gives a strikingly clear picture of the strengths and weaknesses of the program.

From: Franco Andreucci <fran&commat;vmxnucc.cnr,it> Subject: Italian Assistant Software—Automatic translation of my review

This is the automatic translation—done by the Italian Assistant (MicroTac Software)—of the text I posted last week, I didn't intervenc in any word or phrase. Unfortunately, also the texts intentionally written in italian in order to be automatically translated as examples in my review are translated. For instance, if you don't control the original Italian text, you'll miss the meaning of the sentence where L< leader1' is translated with "leader". My criticism was that 'leader' is translated with "duce," Some words arc not translated because the accents are missing, In this case, the responsibility is totally mine.

"Babele, _ W | the Mr. confused the tongue of all the earth'' (Genesis, 11) |by] FRANCO ANDREUCCI

The old man dreams of returning to speak the universal tongue of the Genesis and of annul the chastisement of Babele, hard [e'] to die. In the XIX century he engages the character of the artful idiom and then, in our century, that of the automatic translator. Tied hope a time to the legends of the [positivismo], contradicted from the bankruptcy of the introduction of the [esperanto], she becomes alive anchor in a fascinating and modern way from the protection of the computer. Studied in the Soviet Union in the years '30 and then, after the Second world war, in the United States, the [possibilita*] of the automatic tied translation to the action of a computer has done in the last years of the footsteps from giant. If you/he/shc/it arc thought that the dimen- sion of an electronic dictionary in line [e*] passed from the 250 words of the 1954 to the actual [eentinaiaj of [migliaiaj, we one [puo' j make account that at least a problem |e'J having faced in acceptable way.

Borne from the numerous experiences scientific [svoltesi] in the linguistic field for the auto- matic translation, the idea has found a recent commercial realization in the programs ^Language Assistant Series'η of the MicroTac Software. The programs—that they arc called Italian Assistant, German Assistant, French Assistant, Spanish Assistant and they cost $99,95 each—they foresee the translation in the two senses between the English from a part and the Italian, the French, the German and the Spanish from the other. They represent an enormous footstep in ahead (respect to the by now "Old" dictionaries electronic [tascabili] or to the automatic translators of phrases) for their [elasticita'l and their [capacita'] of answer complex challenges. This critique concerns the Italian part of the program in his release for Windows entirely (MicroTac Software Assistant Windows [for], LverJ. 1.00a).

284 INTERNATIONAL DECISION SUPPORT SYSTEMS

To be able to salvage the message, the user needs to know how to replace the special characters either manually or with special software tools. Hence, the designers of the transnational DSS need to concern themselves with the way in which data are retrieved from corporate databases and transmitted to all users. In addition, designers need to be concerned about the way in which data from external databases, such as network news services, are retrieved and transmitted.

Translations can also affect the user interface in terms of its appearance. One primary problem is the orientation of the text. For example, in the United States, most users feel comfortable with menus that appear at the top of a screen that orient from left to right because that is the way we read. Most standard menuing systems in the United States use such an orientation, and it has been very popular. However, it is common to use a vertical orientation for text in Japan, causing difficulties for software companies trying to make their products more user friendly. It is necessary not only to translate the words in the menus and help screens but also to change the orientation of the entire screen to a vertical framework (associated with their reading and writing conventions).

In addition, many languages are considerably more verbose than English. Or, if the language requires special characters, they may assume more space than standard Roman characters. For example, since Chinese and Japanese characters assume twice the width of a standard Roman character, the standard screen holds only 40 Japanese characters (rather than the standard 80 Roman characters). Hence, translation of elementary aspects of the system design, including prompts and labels, may require an entire screen redesign in order to accommodate the translated terms. For example, consider Figure 6.2, which provides a screen design for a dashboard developed in English, Chinese, Japanese, and Arabic. Notice how the screen needed to be reengineered to accommodate the vertical orientation of the Japanese, the right-to-left orientation of the Arabic, and the range of special characters needed for all three.

Design Insights The Arabic Language

'■ Efforts to develop Arabic DSS have been plagued with problems of how to search for infor- ;

mation in a database. Standard Arabia which is used consistently in written language, has 29 letters, some of which can be adjusted with five different diacritics. In addition, the alphabet consists of several sets of homophones, a rich morphology, and standardized spelling of Arabic names is error prone> Finally, there are almost 20 encodings currently in use for Arabic. Thus, in order to create accurate queries of the database in a DSS, there needs to be some preprocess- ing of the input data. Some have experimented with eliminating the diacritics. Otair, Al-Sardi, and Al-Gialain (2008), however, have developed a more promising intermediary product that attempts to understand the request before transforming them into SQL queries. Their approach processes the words using a stem-based morphological analysis. The tool, called the Arabic Query Analyzer (which is DMBS and application independent), has been fully implemented and has shown tangible performance metrics. A related effort by El-Haj and Hammo (2008) built a query-oriented text summarization system to respond to natural language queries in Arabic* Such a system could help decision makers understand the range of documents, both internal and on the Internet, that might be of help in a choice context. This too has shown promising results.

I I

INTERNATIONAL DECISION SUPPORT SYSTEMS 285

Even when the users can select one language for the system, they may use it quite differently. Researchers in the area of communication long have known that cultures com- municate distinctively.4 Berger (1984, p. 43) notes that "even when they speak the same language, there are problems as a result of differences in education, class, level and cultural backgrounds." Hence, even though the individuals themselves are providing the translations, they may miss the meaning of information, especially if it contains slang or colloquialisms. For example, the British use the term billion to mean what Americans call trillion. That is, the British use thousand million when referring to what Americans call a billion and thus a billion is not encountered until one increases another order of magnitude (hence, the American's trillion). If one were not careful when translating the American version of the English language into the British version of the English language, one might miss the significant implications of size difference.

4"The difference between the almost right word and the right word is really a large matter—'tis the difference between the lightening bug and the lightening" (Mark Twain, U. S. author).

Design Insights The Japanese Language

Japanese text requires special attention in the design of DSS because of the complexity of the language. Some of the issues which contribute to the difficulty for a transnational DSS are highlighted below. In Japanese, one cannot assume that one byte is equivalent to one character, because Japanese characters generally require multiple bytes for representation. The Japanese character set contains over 10,000 characters. The Japanese writing system is a mixture of four different writing systems; Roman, Hiragana, Katakana, and KanjL

■ Roman characters correspond to the 52 characters (including both uppercase and lower- case) of the English language. In addition, there are Roman characters associated with the 10 numerials. Japanese use the Roman characters primarily in the construction of tables and in the creation of acronyms.

* Hiragana characters are ones that represent sounds, such as syllables. Generally, these characters are used to create suffixes for some words or to write native Japanese words. The Hiragana characters appear to have a calligraphic look. For example, the character Ϊ represents the sound made by the letters i(ma" whereas the character = represents the sound made by the combination of letters "mi.71

* Katakana characters represent a phonetic alphabet as well However, they are used to represent words of foreign origin, such as bread, – Prj (pronounced "pan1'), which was derived from the Portuguese word for bread, pdo (pronounced 4tpown1i). In addition, they are used for emphasis, similar to the way we use italics in English. The Katakana characters have a squared, rigid look in comparison to the Hiragana characters. For example, the character ? represents the sound made by the combination of "ma" while the character f represents the sound made by the combination of letters "ku."

* Kanji characters were borrowed from the Chinese over 1500 years ago. There are tens of thousands of these characters in use by the Japanese. These characters represent spe- cific words or combinations of words. For example, * when used alone indicates a tree, while two of the character, **, indicates woods and three of the character, ***, means a forest,

There is no recognized character set for Japanese similar to ASCII for English. Nor is there a universally recognized encoding method for Japanese.

286 INTERNATIONAL DECISION SUPPORT SYSTEMS

(a)

Figure 6.2. Language effects on screen design. The same information is provided in (a) English,

(b) Chinese, (c) Arabic, and (d) Japanese.

Design Insigh nexpected Consequences of Technology Decisions

The move to computerization in cullures with complex alphabets can introduce unwanted impacts on society. Consider the Chinese language, which has roughly 55,000 characters, although only 3500 are in everyday use. When the Public Security Bureau modernized its operations, managers, not surprisi ngly, decided that it would be easier to track i ts citizens if information was computerized rather than handwritten. System designers compromised between the number of characters in ev- eryday use and the census of all characters by allowing the system to use 32,352 unique characters,

While this decision did not have much impact on the operation of the system, or most of the information stored in the system, it did have a major impact on the recording of people's names. Family names were not a problem since only 100 surnames cover 85% of China's 1.3 billion citizens. (By comparison, it takes 70,000 surnames to cover 90% of Americans.) As a result, many Chinese parents look to classical Chinese to find a first name for their children, in part to find a pleasing name and in part to help the child stand out in society. Clearly, these classical names cannot be spelled using the 32,352 characters in the Public Security Bureau's system,

Government officials have told individuals with these unique names that they musi change their name so they can be listed in the database. Further, they are working on a list of "approved1' characters from which future parents must select children's names.

INTERNATIONAL DECISION SUPPORT SYSTEMS 287

(b)

Figure 6.2. (Continued) Language effects on screen design. Translation into Chinese by Aihua Yan.

Translation also needs to be aware of how different cultures adopt the context informa- tion that surrounds the communication. Many cultures of Asia, Latin America, Africa, and the Middle East are high-context cultures. In those cultures, people are highly influenced by the context when interpreting the meaning of communication. So, what is meant depends on the environment in which something is said or written. By contrast, cultures of North America and Australia place more emphasis on what is said to determine meaning than the context in which it is said.

Similarly, different languages and cultures have different ways of representing dates, currency, and other units of measurement. For example, 3/1/10 means March 1, 2010, in the United States, but January 3, 2010, in most of Europe. Many companies in Japan continue to use the Japanese Era Name for years rather than the Common Era designation. So, rather than regarding the year as 2010, they would regard it as Heisei 22 (or 22 years of the reign of the current emperor). Further, some areas of Eastern Asia cling to the "Chinese calendar," which is a blend of the lunar and solar calendars. Similarly, Iran, Afghanistan, and related societies use the Solar Hejri calendar, so the year 2010 would be 1388 or 1389 depending on the time of year (the calendar year begins about March 21 of the Gregorian calendar).

288 INTERNATIONAL DECISION SUPPORT SYSTEMS

(C)

Figure 6.2. (Continued) Language effects on screen design. Translation into Arabic by Michael Martinich-Sauter.

Languages have different rules for pronunciation and therefore meanings which need to be accommodated. For example, a character with an umlaut will have different impacts in Finnish than in German, even though they may look the same to an English audience.

Finally, different languages and different cultures treat the concept of uppercase and lowercase characters differently. For example, the Hebrew language uses lowercase letters only when the text is handwritten and uppercase letters only when the text is printed. In this case, the system designer using a combination of uppercase and lowercase characters in English to convey information would not be able to have the same message sent on the Arabic screen.

Icons can also be a source of confusion when used transnationally because they have quite different interpretations. Those shown in Figure 6.3 are common icons that might be used to give quick visual cues on a dashboard to help the decision maker know whether conditions are improving or not. Clearly, given the range of interpretations of those icons across the world, it would not be prudent to use them in a system that would be used transnationally. In fact, given the internationalization of the employees of most companies, even if they are solely located in a given country, such icons might not convey the intended purpose.

INFORMATION AVAILABILITY STANDARDS 289

(d)

Figure 6.2. (Continued) Language effects on screen design. Translation into Japanese by Mihiro Sasaki.

There is every reason to believe that other less obvious problems of user interface would be different among cultures as well. Unfortunately, if the user interface is unacceptable to users, they will not use the DSS. Hence, it has an important and direct influence on the ability of the user to realize the full potential of the system. The impact of culture upon the database management system and the model management system in transnational DSS is even less intuitive. The remainder of this chapter will highlight some of the legal, cultural, and economic issues that need to be addressed when defining DSS for transnational corporations.

INFORMATION AVAILABILITY STANDARDS

One of the assumptions regarding transnational DSS is that the company can, in fact, share the desired information in all relevant venues. This includes the ability to collect information on a microlevel and to assemble information selectively, to correlate information or in any way create new information from the original data, and to share that information across borders. This implies that the cultures and the laws of the countries are consistent on the view of information, its privacy, and its shareability. In

Related Tags