Skip to content
December 9, 2009 / ABC Translation Services

The Problems With Machine-Based Translations

In a wonderful article by Cecilia Quiroga-Clare, some of the challenges of computer based language translation are discussed. She writes:

“We tend to think of language as a clear and literal vehicle for accurately communicating ideas. But even when we use language literally, misunderstandings arise and meanings shift. People can be intentionally or unintentionally ambiguous. Nevertheless, when someone uses a potentially ambiguous sentence or expression, usually the intention was to express only one meaning. As we know, most words can have denotations, apparent meanings, connotations and implied or hidden meanings. Also, we often use words in a figurative way. Even though figurative language is more often used in poetry and fiction, it is still very common in ordinary speech.

Ambiguity is a poetic vehicle. It is human nature to try to find meaning within an exchange. A text is given to us and in return we give our interpretation. Our own associations give understanding of what is presented to us.

One of the most significant problems in processing natural language is the problem of ambiguity. Most ambiguities escape our notice because we are very good at resolving them using context and our knowledge of the world. But computer systems do not have this knowledge, and consequently do not do a good job of making use of the context.

The problem of ambiguity arises wherever computers try to cope with human language, as when a computer on the Internet retrieves information about alternative meanings of the search terms, meanings that we had no interest in. In machine translation, for a computer it is almost impossible to distinguish between the different meanings of an English word that may be expressed by very different words in the target language. Therefore all attempts to use computers alone to process human language have been frustrated by the computer’s limited ability to deal with polysemy.

The reality is that there no operational computer system capable of determining the intended meanings of words in discourse exists today. Nevertheless, solving the polysemy problem is so important that all efforts will continue. I believe that when we achieve this goal, we will be close to attaining the holy grail of computer science, artificial intelligence. In the meanwhile, there is a lot more to teach computers about contexts and especially linguistic contexts.”

For more information on ABC Translation Services human-touch translations, and why we never use a machine-based translation process, please visit our website at:

%d bloggers like this: