The Conflicting Goals for Machine Translation

Software engineers and developers develop a technology called machine translators. Machine translators are tools that are used by a number of people, ranges from ordinary to professional translations. All of the stakeholders have various goals and wishes for this technology. Despite many translation difficulties, the conflicting goals might be the biggest challenge for the industry. The critical goal of machine translation is not very clear. Should it be the tool that creates ease for translators? Should it work fast to help the clients who deal with multilingual communications all the day? Should it be tuned finely to produce the best quality without any assistance of a man? The latter might be considered a mission impossible in the context of current knowledge. The first is not materialized with the full potential. While the middle is closer to the reality.

The typical machine translation team consists of linguists and technical experts. They are equipped with all the knowledge to introduce technology and manage grammar and languages. But the focus is lost in the pool of linguistic or technical details. The problem can be solved when a machine translates a specific sentence correctly and appropriately.

The fixation process takes time, a lot of time in real. There are many expectations that are needed to be considered. It is very time consuming task to acknowledge and not down all the special cases. The official names and special terms are a nightmare.

The developers have developed a technology that can process a large quantity of text fast, can calculate, can analyse grammar, can divide text into pieces and much more. Does the user want this? He wants to analyse foreign and understand text and get the raw translation to get clear. The user does not need to know about the working of the machine (a cruel truth).

Development of machine translators has different approaches. Practically speaking and statistical machine translators are general approaches, and rule based translators are very slow to use. All the special cases and expectations need to be tracked. Their combination would be worth using but some say it is impossible.

According to professional translators, a customized tool is too technical for quick usage. All types of setups and settings are required. The quality work requires an industry vocabulary as well as terminologies. Wide translation memories take much time to build up. The investment like translation customization needs to return a profit.

Do translators have energy and time to train the machine (tool) or is it easy to post edit? Is it needed to translate from scratch? Post editing and machine translation are built in separate tools that makes it difficult for the translator. Translators get their pay from the outcome. The main goal is to produce a high quality translation on time. The time needed to train the machine or learn about the integration of two separate tools will not produce any money or profit.

As long as language technologies are being developed and users want useful tools, demand/supply do not meet completely in the technology industry. The quality of the translation is not an issue for the unsuccessful market of new translation technology. The user may have complex experience. The machine translation technology may not fulfil the needs or do what the user wants.



Maryum is a freelance writer and IT Professional living in Pakistan. Her technical and transcription blogs about software solutions have been published on various websites. With her hands on experience in translation tools like CATMATE, OMEGAt and TRADOS and advisor of technological tools for big transcription and translation projects. Maryum has the capability of delivering high-quality knowledge in just a few words on her articles and blogs.

Leave a Reply

Your email address will not be published. Required fields are marked *