A machine translator (MT) is a computer that utilizes a program that can translate a document from one language into another language. As businesses become more globally oriented, the need for translation of documents, manuals, contracts and various other texts increases. Human translators are unable to keep up with the demand for their services. In this computerized world, many companies are turning to machine translation to satisfy the growing demand for rapid, high volume translation of technical material of all sorts. This article presents a general overview of the history and development of MT technology and the major types of MTs available in the market and their applications. Finally, the impact of MT, especially in the accounting area, will be discussed.

Through the Years

The idea of machine translation is not new. Initial research into the possibilities of machine translation began in the early 1930s, but it was not until the 1950s that the U.S. government formally recognized MTs and began to support research. A few prototype systems were produced but then abandoned after the National Academy of Sciences dismissed the machines as impractical. Even in the 1960s, research and development of MTs just was not economically feasible, as it hardly made sense to spend millions of dollars to buy a mainframe computer to produce relatively poor translations when a human translator could deliver good ones. Further developments in computer research in the 1970s made the MT a bit more technologically feasible. Although research into this area began slowly, it has recently become a viable alternative, according to the research conducted by Japan and other nations. The MTs of today are economically, technologically and operationally feasible. These new MTs are classified under the umbrella of "artificial intelligence," which aims to reason logically and make judgments as humans do.

Methodology

Basically there are two components in an MT: a dictionary, and information about grammatical rules. Not only does the system examine the words, but it also examines in what context the words are used. The MT analyzes a string of words to identify the meaningful components and the relationships between them, a process known as parsing. Initially, researchers believed that translation could be done by simply matching words by means of a dictionary-based system; for example, how difficult do you think it would be to translate the English word "tree" into the Japanese word "ki?" Look at just some of the possibilities:

Tree

1. A large plant with a woody trunk. 2. A piece or structure of wood for some

special purpose such as a clothes tree. 3. Anything like a tree with its branches,

such as a family tree. 4. A gallows. 5. The cross on which Christ was crucified. 6. A staff, cudgel.

Ki

1. A tree, wood, timber, or lumber. 2. Yellow (color). 3. A season. 4. A vessel, a receptacle. 5. An opportunity. 6. Spirit, mind heart. Not quite that simple after all. Therefore, not only is a dictionary required in an MT system, but also information about grammatical rules. In simple terms, to translate a source language (e.g., English) into a target language (e.g., Japanese), the source sentence must first be parsed, in which a computer will run through all the possible meanings of a word and then try to pick the most appropriate with the help of a table or a list of grammatical rules. Next, this post-parsed, fragmented version of the source language sentence must be reassembled into the target language sentence. Currently there are two ways to do this:

1. The first approach is a direct translation using grammatical and syntactical rules to translate one language into another. 2. The more complicated method involves the use of an intermediate language called "Interlingual." The source language is first translated into the intermediate language. The intermediate language can then be translated into a number of target languages.

The MTs of the 1980s can be run on mainframes, office minicomputers or large microcomputers, as one of two types of translation systems:

1. General Purpose System--such as the Logos Corporation's system that utilizes a standard dictionary of words; and 2. Specialized System--that has a dictionary geared towards specific topics from Accounting to Zoology.

Advantages of Translation Machines

Today, with the expansion of multinational businesses and international trade among countries, MTs have indeed become essential vehicles for expediting important information for multinational corporations (MNCs). The utilization of MTs in processing large amounts of information has increased productivity within firms. Today's MTs are conquering many of the important economic barriers of the 1960s by:

1. Increasing output: from six to eight pages per eight hour day by means of a manual translator, to 18 to 50 pages per day on an MT such as Logos. 2. Increasing translator productivity: 20% to 40% by providing translators with an initial translation to work from. 3. Insuring consistent vocabulary: throughout the entire text. 4. Decreasing total cost: of handling transactions from $55/page to $25/page. MTs cost as much as $500,000 in 1970, today these systems range from $5,000 to $30,000 depending on the number of languages used.

However, the major limitation is that MT is not 100% accurate yet, therefore, the translated outputs must still be post-edited by human translators.

Current Use

Currently, there are systems that translate from Russian to French (Ariane), English to Spanish (Weidner), Chinese to English (Wang), Japanese to English (Bravice International), German to English (Logos), to name just a few. Some are multilingual MTs. Logos is capable of translations between English, German, French, and Spanish. EUROTRA is a system developed under the support of the European Economic Community. This system is intended to perform machine translation among eight languages and is scheduled to be operational by the end of this decade. The applications of these systems will help bridge the language barrier that now exists and will be particularly helpful in the following areas:

1. Government Sector: when governments require that official government documents and publications be made available in several languages; for example, in Canada and Singapore. 2. Business Sector: when businesses have to deal with foreign investors, customers, employees and tourists; for example, in Hawaii and multinational corporations.

Within the business environment, companies have employed MTs to translate large quantities of manuals. For example, Fujitsu of Japan's Atlas Systems translates Japanese computer manuals into English. Translation service companies and hospitals, as well as Honda, have purchased MTs to assist them in their translation needs from Bravice International, a Tokyo software firm. Another company, Hitachi, has also developed a Japanese-to-English translation system for technical documents. In addition to these Japanese firms, Logos Corporation of the U.S. developed a translation system that has been evaluated by the Canadian government for use in translating government documents and publications in English and French. In Utah, Automated Language Processing Systems (ALPS), after years of research and development, now offers software that translates from English into any language that uses the Roman alphabet. Moreover, ALPS is currently working on Japanese, Chinese, and Korean translation software.

Machine Translators and Accounting

Of what use are MTs in accounting? Like the personal computer and fax machines, the MT will probably become a necessary instrument within the accounting industry. As accounting is a service industry, the survival of the accounting firm depends upon its ability to attract and satisfy clients. The MT can aid firm-client relationships and improve productivity and communication within the firm in areas including:

1. Foreign Investment/Joint Ventures. In Hawaii alone, Japanese direct investment reached $1.78 billion in 1988. Across the U.S., joint ventures are becoming an increasingly popular means of doing business as the world economy continues to internationalize itself.

Foreign companies considering investment/joint venture opportunities in the U.S. need advice and support from a CPA firm that can provide service in terms of management information consulting, tax planning and auditing, and in assessing the financial position of the acquisition under consideration. Instead of shying away from foreign clients because of hesitancy towards language barriers, the CPA firm will now be able to tap a whole new source of clientele with the aid of the MT. Rapid translation of timely information could result in prompt decision-making by prospective investors to accept crucial business deals and contribute large amounts of capital to foreign and domestic businesses. 2. Multinational Corporations. MNCs do not exist in a vacuum. Rather, the MNC must make its business decisions based upon the economic and political climate of the country it resides in. Again, the CPA firm can come to the MNC's rescue with assistance in the areas of consulting, tax planning and auditing, as it is very likely that shareholders will be of both countries, and certain that the governments of both countries will require specific types of foreign financial documentation.

With the increase of MNCs, these machines could play a major role in the communication between the parent company and its subsidiaries. The accounting department of the parent company would have an easier time translating documents of its foreign subsidiaries when preparing their financial statements. The tax division of the CPA firm that utilizes the MT will be able to meet the tax planning and preparation needs of the expatriate on a more timely and efficient basis. 3. International or Regional Accounting Organizations. Accounting manuals containing standards, opinions, pronouncements and pertinent information, could be generated via MTs, and made available to countries around the world.

It is hard to foresee the complete impact of MTs in accounting because they are still improving. Overall, these machines should improve communication between countries. The important point is that it is far more productive to have a computer translate a document and a person edit the document than to have a person translate the document from scratch.

David C. Yang, PhD, University of Hawaii