Introduction to Machine Translation
I. Machine Translation (Martínez, 2003; Hutchins, 2003; 2005)
Machine translation (MT) is an automatic translation system that makes use of an
advanced computational linguistic analysis to process source documents and
automatically create target texts with Human-Assisted Machine Translation
(HAMT) or without Fully Automatic Machine Translation (FAMT)1 user
intervention.
It is obvious that most misunderstandings of MT are related to FAMT, as this is the
type of MT that people think of when they hear the term “machine translation”.
However, this automatic translation can be a reality only under very strictly defined
conditions where the language and style of the input is restricted as a consequence of
dealing with very specific domains (e.g. technical), or when MT output, although (very)
low in quality, is accepted.
II. Causes of MT Errors (Hutchins, 2003; 2005)
The reasons for MT mistranslation may be relatively trivial, but the main MT errors
arise from the following:
• The difficulties computers have with many aspects of language (ellipsis2,
pronouns, coordination, etc.) and with handling complex sentences – long
sentences of more than one clause tend always to be translated less successfully
than short single-clause sentences.
• If the original text includes a misspelling, the word will not be recognized,
and this will affect translation of the rest of the sentence.
• If the original contains a typographic mistake (e.g. the use of from instead of
form), there is going to be a problem with translation.
• Prepositions are always problematic for MT systems, and prepositional verb
phrases or phrasal verbs are equally problematic.
1
The computer process of translating a source text (ST) into a target text (TT) without any human
intervention.
2
The omission from speech or writing of a word or words that are superfluous or able to be understood
from contextual clues
1
I. Machine Translation (Martínez, 2003; Hutchins, 2003; 2005)
Machine translation (MT) is an automatic translation system that makes use of an
advanced computational linguistic analysis to process source documents and
automatically create target texts with Human-Assisted Machine Translation
(HAMT) or without Fully Automatic Machine Translation (FAMT)1 user
intervention.
It is obvious that most misunderstandings of MT are related to FAMT, as this is the
type of MT that people think of when they hear the term “machine translation”.
However, this automatic translation can be a reality only under very strictly defined
conditions where the language and style of the input is restricted as a consequence of
dealing with very specific domains (e.g. technical), or when MT output, although (very)
low in quality, is accepted.
II. Causes of MT Errors (Hutchins, 2003; 2005)
The reasons for MT mistranslation may be relatively trivial, but the main MT errors
arise from the following:
• The difficulties computers have with many aspects of language (ellipsis2,
pronouns, coordination, etc.) and with handling complex sentences – long
sentences of more than one clause tend always to be translated less successfully
than short single-clause sentences.
• If the original text includes a misspelling, the word will not be recognized,
and this will affect translation of the rest of the sentence.
• If the original contains a typographic mistake (e.g. the use of from instead of
form), there is going to be a problem with translation.
• Prepositions are always problematic for MT systems, and prepositional verb
phrases or phrasal verbs are equally problematic.
1
The computer process of translating a source text (ST) into a target text (TT) without any human
intervention.
2
The omission from speech or writing of a word or words that are superfluous or able to be understood
from contextual clues
1
