Iste presentation se divide in quatro partes: (1)
Le synergia inter systemas electronic de traduction
e traductores human, (2) Le importantia del
accessibilitate rapide de dictionarios electronic,
(3) Le uso efficace del systemas currente, e (4)
Systemas del futuro.
---
(1) Le synergia inter systemas electronic de
traduction
e traductores human:
Io crede que io ha demonstrate que on pote producer
traductiones assatis sophisticate con
<http://www.systransoft.com> in su forma currente.
Il es evidente que on nunc continua (sta a continuar)
le travalio de disveloppar iste systema de traduction,
addente nove linguas a illo e meliorante le
capacitates
del logiciel pro omne le linguas que forma parte
del systema.
Systemas electronic de traduction nunc pare esser
assatis potente que illos pote adder significantemente
al productivitate de traductores human. Un bon
maniera de usar los esserea passar textos que on
vole traducer trans le traductor electronic e facer
apparer e le texto original e le traduction preliminar
sur un schermo in un formato bicolumnar. Traductores
tunc poterea comparar le duo textos e continuar
le redaction del texto final.
(2) Le importantia del accessibilitate rapide de
dictionarios electronic:
In lor lor effortios pro trovar equivalentes plus
exacte inter le duo textos, le traductores human
anque deberea poter accessar rapidemente dictionarios
electronic bilingue in un fenestra sur le schermo.
Tal dictionarios bilingue esserea plus utile si
illos provideva equivalentes de unitates semantic
in le forma de phrases e de asseverationes. In
iste momento illos non ha ulle cognoscimentos
morphosyntactic. Con le tempore, il essera possibile
construer los con iste capacitates a un nivello
basic, e eventualmente illos poterea haber unes
ex le capacitates de systemas dedicate al traduction
de textos longe como systransoft.
(3) Le uso efficace del systemas currente:
In su formas currente programmas de traduction
electronic como systransoft pote producer resultatos
melior si on non los usa naivemente. On debe
cognoscer
un pauco le limites de tal systemas e cognoscer
ambe le linguas con le quales on travalia durante
que on los usa.
Io, como anglophono native, naturalmente formularea
in anglese omne le textos que io volerea traducer.
(Le nativos de omne linguas ha un flexibilitate
de expression attingibile per un numero multo reducite
de personas qui los apprende como secunde, tertie,
o quarte linguas--specialmente si tal studentes
comencia lor studios post le etate de unes dece-duo
o dece-tres annos.)
Quando io redige mi textos in anglese ante introducer
los a un traductor electronic, io essaya usar le
vocabulario le plus frequente e le structuras le
plus simple del anglese.
Pro exemplo, in vice de scriber "He gave all these
efforts up" io essayarea scriber "He gave up all
these efforts". Si le traductor electronic non
poteva travaliar con le asseveration in iste forma,
io usarea "He abandoned all these efforts" (ille
abandonava omne iste effortios).
In su forma currente, quando Systransoft produce
textos in espaniol, illo scribe "la idioma" in vice
de "el idioma". Si io non habeva studiate el
espaniol,
naturalmente, io non haberea recognoscite le
necessitate de cambiar le articulo "la" a "el" in
iste contexto. Io es confidente tamen que in le
futuro on corrigera iste e altere simile problemas.
(4) Systemas del futuro:
Possibilemente in le futuro, quando nostre
computatores
devenira etiam plus rapide e intelligente con le
appoio de quantitates de memoria multo plus grande,
il essera possibile construer systemas pro travaliar
con le linguage que functiona de un maniera que
resimila le processos del evolution biologic.
Iste systemas studiarea le travalio del redactores
del textos final e lo incorporarea automaticamente in
lor proprie algorithmos de traduction. On jam ha
construite un tal systema in Egypto pro traducer inter
le anglese e le arabe, le qual ha reducite de un
maniera significante le travalio del traductores
redigente le textos final que illo produce.
Un systema que imitarea le processo del evolution
biologic tamen esserea multo plus flexibile in su
capacitate que le motores de intelligentia artificial
currente. Sin dubita, un tal systema poterea producer
le resultatos del systemas currente solmente con
quantitates multo plus grande de processamento e
de memoria. Ma con le celeritate plus grande del
systemas electronic del futuro, iste programmas
de traduction probabilemente facera lor travalio
al minus tan rapidemente como systemas currente--e
forsan etiam plus rapidemente!
...............
This presentation is divided into four parts: (1)
Synergy between electronic translation systems and
human translators, (2) The importance of rapidly
accessing electronic dictionaries, (3) The efficient
use of current systems, and (4) Future systems.
---
(1) Synergy Between Electronic Translation Systems
amd Human Translators:
I think I have shown that it is possible to produce
rather sophisticated translations with
<http://www.systransoft.com> in its current form.
It is evident that developmental work is continuing
on this translation system with the addition of
new languages and with improvements in its software
capabilities.
Machine-tanslation systems now seem to be powerful
enough to add significantly to the productivity
of human translators. A good way of using them
would be to run texts texts to be translated (source
texts) through the computerized translation system and
displaying them and the preliminary translated texts
on a screen in a two-column format. Translators then
could compare the two texts and continue editing the
final text (target text).
(2) The Importance of Rapidly Accessing Electronic
Dictionaries:
In their efforts to find more exact equivalents
between the source and target texts, human translators
should also be able to quickly access electronic
bilingual dictionaries in a window on the screen.
Such bilingual dictionaries would be more useful
if they provided equivalents of units of meaning
in the form of phrases and sentences. Right now
they don't have any morphosyntactic knowledge (in
other words, they wouldn't know that "this books"
should be "these books). In time it will be possible
to build into them this kind of capacity up to a
basic level, and eventually they could have some
of the capabilities of systems like systransoft
that were built to handle long texts.
(3) The Effective Use of Current Systems:
In their current forms machine-translation systems
like systansoft can produce better results if they
aren't used naively. A translator should be somewhat
aware of the limits of such systems and know something
about both languages while using them.
I myself, as a native speaker of English, would
naturally use English to produce all my source texts.
(Native users of all languages have a flexibility
of expression attainable by only a very small number
of people who learn them as second, third, or fourth
languages--especially if such people start their
studies after the age of about twelve or thirteen.)
When I edit my English texts before feeding them
through an electronic translator, I try to use the
most frequently occurring vocabulary and the simplest
structures of English.
For example, instead of writing "he gave all these
efforts up" I would try to write "He gave up all
these efforts." If the electronic system could
not work with the statement in this form, I would
use "He abandoned all these efforts."
In its current form, when Systransoft produces texts
in Spanish, it writes "la idioma" instead of "el
idioma." If I had not studied Spanish, naturally,
I would not have recognized the need to change the
article "la" to "el" in this context. I am confident,
however, that sometime later on these problems will
be corrected.
(4) Future Systems:
Possibly in the future when our computers become
even faster and more intelligent and are supported
by much more memory than right now, it will be
possible
to build systems with language-processing capabilities
resembling the processes of biological evolution.
These machines would study the work of the people
editing the final texts and automatically incorporate
them into their translation algorithms. Such a system
has already been built in Egypt to translate between
English and Arabic and has reduced significantly the
work of the editors working on the final texts that it
produces.
A system that would imitate the processes of
biological
evolution, however, would be much more flexible
in its capacity than current artificial-intelligence
engines. It would no doubt be able to achieve the
results of current systems only with a lot more
memory and a lot more processing. But with the
greater speed of future electronic systems, such
translation software would probably work just as
fast as what is currently available, and maybe even
faster!
__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/
|
