| 421.1.9.3 : FORMAL EQUIVALENCE |
|
| It was made
clear at the beginning of this section in translating
we are concerned not only with finding suitable equivalents
of lexical items between SL and TL, but also interested
in the question of how the formal patterns of the two
languages compare. In this regard we should remember Nida
and Taber's (1974:3) caution that "each language has its
own genius". The word order patterns in a language, the
device for linking clauses into sentences or the manner
in which discourse units are marked in a language all
give a language, its distinct character. Successful translation
consists in recognizing and giving due credit to the genius
of the language in question. As Nida and Taber (1974:3)
insist "rather than bemoan the lack of some feature in
a language, one must respect the features of the receptor
language and exploit the potentialities of the language
to the greatest possible extent." |
|
| But while
emphasizing on the one hand, the unique structural properties
of a language, Nda and Taber also suggest that there are
in all languages a limited set of basic or 'kernel structures'.
It is out of these kernel structures that more elaborate
structures are formed by means of translation. For instance,
a kernel structure of the type John hit Bill can be taken
as underlying the structure of a sentence as Bill was
hit by John the latter being derived from the former by
the passive transformation. Nida and Taber (9174:39) pointed
out that from the standpoint of the translator what should
be important is not the existence of 'kernels' in all
languages, but "the fact that languages agree far more
on the level of the kernels than on the level of the more
elaborate structures". In fact, they suggest that in translation
if grammatical structures could be reduced to the level
of kernel the transference from one text to the other
will involve a minimum of distortions. |
|
| The search
for a universal base vocabulary and syntactic structures,
however, has gained more prominence in Machine and computer
aided translation. Of the different approaches to Machine
Translation it is the 'interlingual approach' in which
the need for a common or universal semantic and structural
representation become quite pronounced. As different from
a 'direct translation' approach in which we compare the
vocabulary and syntax of SL and TL to find out the equivalents,
in the interlingual approach a different method is adopted.
It is assumed that it is possible to convert the SL text
into semantic and syntactic representations which will
be common to more than one language. Translation of a
text in this approach thus proceeds in two stages. The
first stage consists of analyzing SL texts into interlingual
forms are synthesized to give texts in the TL. In the
interlingual approach the 'interlingual language' itself
can be conceived of differently. For instance, some tried
creating an 'interlingua' which would be a direct representation
of conceptual relations independent of any language. Yet
another way of looking at interlingua was that if could
be the sum of all features to a set of languages taken
up for translation. A third alternative proposed is that
the interlingual could be a complete artificial language
having its own lexicon and syntax created on the basis
of the most common features of the languages in question.
But what is common to all these alternatives is an attempt
to arrive at a common or universal base or interlingual
which will facilitate intertranslatability. |
|
| It should
be clear from the foregoing sections that for carrying
out translation from one language into another translation
theory looks for a common base both at the level of vocabulary
as well as at other levels of language organization. Nida
and Taber's proposal of 'kernal structures' and 'interlingual'
of machine translation are instances of this common base.
We can therefore conclude that universals of language
are relevant not only for theoretical linguistics but
also in different applied areas including translation. |
|
| 421.1.9.4: CONCLUSIONS |
|
| However
no claim is being made here that unearthing of deep rooted
similarities among languages would lead us automatically
to set of formulate with the help of which texts of one
language could be automatically converted into those of
another. Two languages may agree in having similar structural
patterns but the languages may not employ these structures
for parallel functions. Let us consider one example. Both
English and Hindi have a syntactic construction called
coordination. The two clauses in question are coordinated
by and in English and by our in Hindi. But an English
sentence like she came and game the book cannot be translated
by a coordinate structure in Hindi. That is, to translate
the above sentence into Hindi as vah aayi aur usne mujhe
kitab dii 'she came and she gave me the book' will be
awkward. Such English sentences can be better translated
into Hindi and other Indian languages by a participle
construction such as usne aakar mujhe kitaab dii. Conversely
two languages may differ structurally but may allow better
scope for translatability due to their socio-cultural
affinity. For instance, though Indian languages belong
to a number of different language families, yet, they
share a lot easy translatability among them. In short,
establishing equivalents between the source and target
language itself does not ensure good translation. |
|
| It has generally
been felt (Gross, 1992) that the development of computational
linguistics and computer aided translation is a direct
result of the generative approach to linguistic universals
and universal grammar. This universalist approach has
been criticized for indulging in stunning generalizations
by ignoring the study of individual languages and cultures.
But it should be noted here that though Chomsky is a potoganist
of linguistic universals, he has not advocated that discovering
universal properties of languages will enable us to translate
mechanically from one language into another. Let us consider
Chomsky's own opinions on the subject (1965:30). "The
existence of deep seated formal universals implies that
all languages are cut to the same pattern, but does not
imply that there is any point by point correspondence
between particular languges. It does not, for example,
imply that there must be some reasonable procedure for
translating between languages". What does Chomsky mean
by 'reasonable procedure'? Chomsky says "by a "reasonable
procedure I mean one that does not involve extra-linguistic
information. That is one that does not incorporate an
"encyclopedia",) 1965:202, footnote 17)". Contrary to
the criticism stand referred to above, Chomsky has adopted
a very cautious stand regarding the role of language universals
and translatability. In fact Chomsky is even skeptical
as to whether machine translation is feasible. At one
point he says (1975:40) " as for machine translation and
related enterprise, they seemed to me pointless as well
as probably quite hopeless'. Perhaps, the truth lies somewhere
in the middle. We cannot rule out completely the possibility
of utilizing the underlying commonalities among languages
for producing texts from one language into the other to
cope with the massive explosion of information in the
world today. We should remember at the same time that
language is an organized formal system which is used by
a speech community for expressing its socio-cultural ethos
in a creative manner. This creativity of language use
cannot be reduced to a set of mechanical rules. |
|
| Points
to Remember |
|
|
