Monday, 12 August 2013

Computational linguistics


Computational linguistics is the study and use of computer programmes which analyse language in automatic translation or properties of texts.

The first automatic calculation machine was designed by Babbage in 1822. However, the first programmable computer was invented just after World War II. After this, computer scientists in America began to develop applications to translate documents. Particularly, to translate research papers regarding the Apollo-Soyuz space project from Russian to English.
Even though computers use arithmetic calculations in their algorithms, machine translations were unable to produce accurate translations. It was discovered that in order to translate languages using computers they had to factor in the morphology, syntax, lexicon, semantics and pragmatics of both languages. This was a very large ask and an accurate translation software has still not been created.
 Currently, the most advanced computers are able to translate 10,000 words per hour. Although, afterwards translations need further editing to make them accurate and coherent, the computational method is much faster than translators working ‘by hand’.
However, it would be naïve to say that translation is the only way computers can help linguists. For example, computers allow linguists to gather large amounts of discourse recordings and analyses them automatically. This technique can be used in theoretical computational linguistics to form hypotheses and predictions about language and behaviour.
Scientists are always trying to improve current translation and analysis programmes but it is unlikely that machines will completely replace human translators or linguists in the near future despite their efforts.

 
More information is available at:
Page 360-361 in 'The Encyclopaedia of Language' by David Crystal

No comments:

Post a Comment