Ïåäàãîãè÷åñêèå íàóêè/5. Ñîâðåìåííûå ìåòîäû
ïðåïîäàâàíèÿ
Kostanai State
University named after A. Baitursynov, Kazakhstan
CURRENT
PROBLEMS OF MACHINE TRANSLATION
The advent of
computers and information technologies has created a new electronic culture
which replaced the print culture and the process of working with documents and
information has been greatly simplified and sped. Machine translation programs
and computer translation tools have become part of this new culture, so that
this refers to a machine and automated translation. [4]
These concepts
shouldn’t be confused. Very often, instead of "machine" it is used
the word "automatic" and it does not affect the meaning. However, the
term "automatic translation" has a very different meaning. Machine
translation is the process of translating texts from one language to another
using a special computer program, implemented by this program independently,
albeit with subsequent editing of the translator. The automated translation
programs simply help a person to translate the text. They are just tools –
dictionaries, databases, glossaries facilitating and accelerating the
traditional process of translation. Automation has the purpose of facilitating
the work of the person, substituting of his hand forms and increasing
productivity, while machine translation technology is used in order to minimize
human involvement in the process, and ideally remove it from the process. [1] Yet
still it is not achievable at this stage of technology development. So far
there are no programs that are able to give completely adequate translation of
foreign texts. Therefore, the problem of machine translation remains relevant
at the moment.
Modern technology has gone far ahead of the first attempts to
"stick a translator into computer." FAMT- (Fully-Automated Machine
Translation) and HAMT- (Human-Assisted Machine Translation) -systems are a
footstep away from being called as artificial intelligence systems, as they
already perform certain functions of the human brain: in particular, they
design the text on the object data language based on the source one using the
set of certain rules given in the form of structures and algorithms. There
exist no absolutely universal algorithms, of course, for FAMT- and HAMT- nor
even for MAHT (Machine-Assisted Human Translation) -systems, as different
products are based on different approaches to translation. Nevertheless a
general scheme can be made. It is worthy of note that this scheme is
approximate, simplified, and within each procedure both human and machine carry
out a range of different actions. A man enters the text into the computer and
performs manual setting of SMT, i.e. defines the parameters of the source and
object language, selects the base theme and terminological dictionaries for
translating specialized vocabulary, sets limits for translating names appearing
in the text, etc. The program searches for the word forms in the dictionary of
the source language, and performs the morphological analysis of the input
information. During the analysis other information may be obtained. The program
searches for matches. First, it is made the translation of idioms,
phraseological unities and stamps in this domain. Then they perform the
definition of the basic grammatical characteristics of the remaining elements
of the input text (e.g. nouns, verbs, tense, etc.). Usually at this stage
monosemantic words are separated from the polysemantic ones, then the monosemantic
words are translated according to the lists of equivalents, and for the
translation of polysemantic words are used specialized dictionaries. Lexical
analysis of the input information and the actual translation itself complete
the procedure. At this stage people may get in the work of some programs to
“prompt” the machine extra linguistic nuances that are not clear for it: e. g.
which of the several meanings of the word should be chosen in this case.
Then the program
performs a grammatical analysis of the resulting draft translation, in the
course of which it is defined the missing grammatical information taking into
account the data of the object language (for example, it is turned out, which
of the three tenses of Russian verb in this context corresponds better to the
"present continuous" tense of the English verb). It is performed the
synthesis of output word forms and of the text as a whole. At this stage the
work is done again by people who correct mistakes of the machine translation.
Thus, machine translation
and computer-aided tools are instruments that can make the translation process
more efficient, i.e. faster and less expensive.
But if computer
tools really speed up the translation process, the efficiency of machine
translation systems up to date is still questionable, despite the presence of
advanced modern technologies. Therefore a translation done by a machine
translation program, in any case cannot be regarded as a "finished
product", it needs serious improvement. And despite the fact that the
machines are made by people they cannot think like a man. Although the machine
translation program provides the translated text, but errors in this text
differ from the mistakes that people make. Besides, in this case, the approach
to translation changes itself – the translator does not translate, but corrects
mistakes, and mistakes are not human, these are mistakes of the machine. If we
develop this idea, we can conclude that the work with a machine translation
program does not need a translator, because a man who knows both these languages can do this work. [4] But it is difficult
to agree to this point of view, because if a person owns at least one foreign
language, he automatically owns some translation skills, therefore, is a
translator in substance. Besides, it is doubtful that a person who has no
linguistic knowledge could competently correct errors of the machine.
Therefore, in the
training foreign languages one of the skills that students must master is the
ability to perform the automatic translation of the text accompanying by his editing skills.
Literature:
1. Solovyov, A. V. Professional
translation by computer. [Text] / A. Solovyov. – St. Petersburg.: Peter, 2008.
– 160s.
2. Retsker, L. Translation theory
and practice of translation. [Text]: Essays in linguistic theory of translation
/ L. Retsker. – M., Valent, 2004. – 244.
3. Machine
translation [electronic resource] // URL: http://www.baidak.com/blog/machine-translation
4. All about
machine translation [electronic resource] / / All about computers – URL:
http://www.computerbild.ru