| AUTHOR: Uwajeh, M.K.C.
TITLE: Translation Equivalence
SUBTITLE: An Essay in Theoretical Linguistics
SERIES TITLE: LINCOM Studies in Translation 01
PUBLISHER: Lincom GmbH
Anil Kumar Singh, Language Technologies Research Centre, International Institute
of Information Technology, Hyderabad, India
The book presents a theory of translation and emphatically situates the
scientific study of translation within the domain of Linguistics as a Science
(Fromkin, 1999). The author says that this is in opposition to the more popular
view of translatology in which Linguistics is considered of a peripheral
importance. In fact, the subtitle of the book is ''An Essay in Theoretical
Linguistics''. Although the proposed theory is mainly in consonance with the
central ideas of modern Linguistics such as the competence and performance, it
is based more specifically on what is called 'Performative Linguistics' (Uwajeh,
1994). The author views translation as a four level exercise: lexical, literal,
free and figurative. The objective in translation, according to the author, is
to achieve 'equivalence', i.e., whatever is conveyed during translation should
be ''of equal value'' rather than ''same in meaning''.
The book is divided into seven chapters. It starts with a short introduction and
ends with a short conclusion. The content part of book is only a little more
than two hundred pages long. At the beginning, before the preface, a neat
diagram outlines the ideas described in book.
In the Introduction, the author provides some background to the content of the
book. He mentions that the book is designed especially for ''students of
translation'', irrespective of whether their interest in translation is ''central''
or ''incidental''. He says that he has tried to make the book ''as clear and simple
as possible by telling directly my own story of what translation is''. The result
is a textbook on translation, but ''not in the traditional sense''. He also
emphatically states that the theory presented in the book is categorically a
linguistic approach to translation, i.e., ''any viable theory of translation is
part of Language theory''.
The first chapter starts from the idea that translation is essentially is
linguistic operation such that the translation activity is a type of language
The author argues for the position that language is essentially a communication
tool. In any instance of language use, two parties are involved, one of them is
the information-sender (communicator), while the other is the information
receiver (communicatee). The author compares language to aircraft: A language is
meant to communicate just as an aircraft is meant to fly. Both have a specific
However, communication can be 'representation' or 'indicational'. Language is a
representational means of communication, while other communication tools are
indicational means of communication. Indicational communication is more general
than representational and includes the latter. Such characterization of language
follows from what is called Performative Linguistics (Uwajeh, 1994), a context
sensitive and performance approach to the scientific characterization of
language started by the author and others at the Universit Montral in the late
Representation, the author argues, is a specialized means of communication where
some particular state of affairs is ''knowable in advance'' by the communication
parties involved. The communication intent ''a'' is 'stated' by ''b'', not
'implied'. The ''a'' is 're-presented' by ''b''. Given some information, the
representer 'recalls' ''a'' and the communicatee 'recognizes' that ''b'' represents
This is followed by discussions of intralinguistic and extralinguistic
communication skills. The intralinguistic skills concern the language 'texture'
and are the core constituents of language communication skills. According to
Performative Linguistics, language may be defined as ''any semantic-symbolic
(i.e., thought and symbolization or meaning-and-form) potentiality'' or
''intercommunication representational system structure''. Intralinguistic
communication can also be defined as ''the ability to construct thought
structures as well as their corresponding representing symbolization structures
for the communication of experiences, and the ability to comprehend what is
communicated with those two structures''.
The author then presents, (i) ''the three Rs of language construction'', and (ii)
''the three Rs of language comprehension''. In (i), R1 is the Experience or the
Referent, R2 is Thought or the Reference and R3 is the Symbolization or the
Referend. Here R2 and R3 are parts of the language, with the former representing
the semantic level and the latter representing the symbolic level. On the other
hand, in (ii), R3 is the Form or the Referend, R2 is the Meaning or the
Reference and R1 is the Information of the Referent.
In the next section, the author emphasizes the distinctive nature of
representational relationships. A representational relationship does not
necessarily imply any resemblance between the representer and the represented.
The point that the author makes repeatedly is that as opposed to the 'arbitrary'
nature of this relationship as is expressed by many linguistic scholars, it is
actually ''irrelevant'' whether there is any resemblance or not. For example,
there is indeed some resemblance in the case of onomatopeic words.
The author then goes on to differentiate between active and passive forms of
language competence. The former refers to the communicator's ability to produce
texts of language, while the latter refers to the ability to comprehend texts
produced for him/her to interpret. Our active competence is, in general, weaker
than our passive competence. This point is relevant from the translation point
of view because all translators should possess very good active competence of
their working target languages. Note that this view of competence is different
from the Chomskian view of competence (Chomsky, 1965) and is similar to the view
of some other scholars (Savignon, 1997). The former defines competence as the
knowledge of rules and structure, whereas the latter defines it as the ability
to use language for some purpose.
The 'extralinguistic skills' are the skills of language context. They comprise
what it entails to know how to use language for that use to be appropriate. They
are the peripheral constituents of language communication skills. They may also
be described as the modalities or conditions for communication with language to
be ''suitable''. These skills themselves can be categorized as general or
specific. The general extralinguistic skills concern the knowledge of world
realities: Concrete or material (trees, houses); abstract or mental (beliefs,
values); and process or behavioral (painting, singing). The specific skills, on
the other hand concern the knowledge or those contextual factors that are the
typical constraints on the communicator for specific language communication acts
and comprise the actual (typical) circumstances for the language-using
individual's language choices in communication as an integral member of a
particular language community.
In Chapter 2, the author first outlines the Translation Communication Process,
discusses different types of translations and then presents Uwajeh's Four Levels
The Translation Communication Process (as I have understood it) can be outlined
as follows. The sender (original communicator) starts with the realities
(experiences or the referent) on the source side. The input in the source
language comprises of the reference A and the referend A. This is filtered
through the translator, who is the stand-in communicatee on the source side and
the stand-in communicatee on target side. The result comprises the realities
(experiences or the referent) on the target side. The output in the target
language comprises the reference B and the referend B. At the end, the output is
received by the receiver (original communicatee). The 'referent' here means the
realities, or the information conveyed with language. The 'reference' means the
thought, or 'language meaning'. And the 'referend' means the symbolization, or
the language form.
It is emphasized that translation is a one-way communication activity in which
the translator acts as a kind of filter, bridging the communication gap. Also,
translation is not a transfer of anything. Rather, ''the same'' information
conveyed with the given language (SL) is reconveyed with another language (TL).
The primary preoccupation of language use known as translation is information,
not meaning as such because meaning is not constant across languages. Finally,
there are actually two communications in translation, as is implied by the
description above. This view of translation might be seen as different from the
'transfer based' translation, an idea that was once quite popular in the machine
translation community (Hutchins & Somers, 1992, Arnold et al., 1993). However,
it can be argued that the meaning of 'transfer' in these two cases is not the same.
The author then gives a somewhat long description of how translation can be
categorized. The criteria used for classification are 'time' (retroactive,
immediate, consecutive and simultaneous), 'medium' (written, oral or gestural),
'subject matter' (overtranslation, undertranslation, literary, scientific etc.),
'textuality' (full, partial) and 'agency' (human, machine, machine-assisted,
automatic). The author points out that true machine translation is still in the
realm of science fiction.
The second chapter ends with the presentation of the four levels of translation.
The central problem of translation is 'equivalence' (Catford, 1965) or 'of equal
value' and the central basis for equivalence in translation between target
language textures and source language textures is information. There are four
standard types of equivalences applicable for SL-TL translations. These are:
* Conceptual equivalence (sameness of concept units)
* Propositional equivalence (sameness of thought patterns)
* Thematic equivalence (sameness of subject-matter)
* Contextual equivalence (sameness of context variables)
Corresponding to these levels of equivalence, there are four levels of translation:
* Lexical translation: A lexical-item-by-lexical-item translation
* Literal translation: A rough-rendition synthetic translation, emphasizing the
''global meaning structure''
* Free translation: A smooth flow translation
* Figurative translation: A ''special effects'' translation (where applicable)
Each of the above translation levels targets a specific 'degree of equivalence'.
The first two SL-oriented translations while the last two are TL-oriented
translations. Note that such view of translation is different from the
conventional three level based theories of translation.
Here is an example that illustrates the four levels of translation:
Tínyé áka nà àkpà-á àkàkpò, kà íwèlí ífé li nà àkpà-á àkàkpò.
Tínyé áka nà àkpà----á àkàkpò, kà
[PIT-IN HAND in POCKET OF SHORTY, that]
í------wè---lí ífé li nà àkpà---á àkàkpò.
[YOU TAKE UP THING is in POCKET OF SHORTY.]
[[Put you hand in the pocket of shorty, so that you
take up the thing that is in the pocket of the shorty.]]
(Dip your hand into the shorty's pocket and take out what is in shorty's pocket.)
((Peter Piper picked a peck of pickled pepper.))
Chapter 3 is about conceptual equivalence. It is the first level of semantic
equivalence, and is also the first level of intralinguistic equivalence. Such
equivalence achieved at the level of lexical translation, which is not just
word-for-word translation. This is essentially a meaning, not form affair. In
the author's terminology, a 'lexical item' (de Saussure, 1916) is a fundamental
linguistic unit comprising a basic unit of thought or 'meaning' and a basic unit
of symbolization or 'form' which represents the conceptual unit. This is
different from the conventional usage where words generally understood to be
only 'forms'. Thus, the conceptual equivalence is achieved through
The author states that conceptual equivalence is almost certainly the hardest to
achieve in practice, even if it is not necessarily the most difficult to attain
in essence nor the most important in translation as a whole. This is due to the
''extent of current ignorance about concepts'', i.e., the intralinguistic problem
of 'conceptual indeterminacy' (concepts are largely indeterminate) and the
crosslinguistic problem of 'conceptual inconstancy' between SLs and TLs. The
author also discusses how these problems can be overcome in the practice of
In Chapter 4, the author explains propositional equivalence in more detail.
'Proposition' is any thought unit larger than the concept. Thus, propositions
are made up of concepts. The purpose of propositional equivalence in translation
is to present as faithfully as possible with the target language texture for any
member(s) of the TL community interested the thought pattern of any member(s) of
the SL community who used source language texture to communicate. Here, the
author argues for a literal level of translation in addition to a lexical level
of translation, which is why the author's model has four levels as opposed to
the conventional three level models. The author cites the examples of _Nineteen
Eight-Four_ (Orwell, 1949) and _Things Fall Apart_ (Achebe, 1958). To these, one
might add _For Whom the Bell Tolls_ (Hemingway, 1941) in which the English used
by characters follows, in Uwajeh's terminology, Spanish 'texture'. Literal
translation tries to preserve the SL community's 'world picture'.
An example from _For Whom the Bell Tolls_ which seems to validate Uwajeh's
contention is given below:
''Perhaps twenty. Depending how many they would bring for this business. If they
would come for this business. Remember thee that in this of a bridge there is no
money and no loot and in thy reservations of talking, much danger, and that
afterwards there must be a moving from these mountains. Many will oppose this of
In a section titled ''Sapir-Whorf Hypothesis Revisited'', the author clarifies his
position with regard to 'linguistic relativism' (Sapir, 1947, Whorf, 1956). He
says the correct position corresponds neither to the weak nor to the strong
version of the Sapir-Whorf Hypothesis. He refers back to the 'three R's' of
language construction and language comprehension and says that the relationship
between language thought and reality is 'neutral' like the relationship between
language thought and language symbolization. In other words, it is irrelevant
whether there is any similarity or resemblance between these two pairs or not.
Later in the chapter the author says that the function of literal translation is
to expose with TL any oddity of the SL community's world picture noticed from
the TL community's world picture. Thus, in literal translation, the thought
patterns are from the SL, but the grammar is that of the TL.
In Chapter 5, the author elaborates on the thematic equivalence. By 'theme', the
author refers to what is being communicated about (directly) with a language.
They are the material for the information stated with both source language
textures and target language textures. They are the 'realities' of the language
community as a social network, representable for communication the language
'texture'. The task of the translator is to achieve equivalence in their
representation between SL and TL.
Thematic equivalence concerns the first level of TL-oriented translations, and
also the first level of extralinguistic equivalence. This level of translation
is what the author calls 'free translation' and it is the most sought after
level of translation. The basic function of thematic equivalence is to report
directly with target language texture as faithfully as possible for any
member(s) of the TL community interested in the same reality as that reported
directly with source language texture for any member(s) of the SL community.
The author states that one major problem for achieving thematic equivalence is
that free translations are often wrongly designated as literal translations.
This, the author says, happens because of a considerable latitude in the usage
of the expression 'literal translation'.
Just like the previous level, thematic equivalence is hard to achieve because of
'thematic inconstancy' between the SL community and the TL community. At its
extreme, it might render correct translation impossible. The author gives the
example of Newspeak (Orwell, 1949) where the members of the Newspeak community
are simply incapable of understanding the text of Oldspeak, even though, one
might say, the language is still technically the same (English). Therefore, for
thematic equivalence, at least some common background of social realities is
indispensable for communication. The author cites this as one of the reasons why
true machine translation is likely to ''remain in the realm of science fiction'':
Because the machines lack extralinguistic knowledge.
Chapter 6 covers the fourth and highest level of translation equivalence, namely
contextual equivalence. The context here means immediate 'psychological context'
of language communication. The author proposes that there are seven types of
factors (called 'variables') which determine the communication context. These
are the seven wh-words: who, where, what, why, when, which, and whom.
This level of translation is the what is commonly called the 'figurative
translation'. The difference between this and the previous level (according to
the author) is that the information is communicated indirectly, i.e., indicated,
not (directly) represented. An example that the author gives is the two
different translations of 'Good morning' in the Kwale or Ukwuani dialect of the
Igbo language which is spoken in a part of Nigeria: 'Bather' if addressing a
grown up unmarried female, and 'Soup maker' if addressing a married female.
The function of contextual equivalence is to communicate as faithfully as
possible with a target language texture for any member(s) of the TL community
concerned about the same communication context whereby the source language
texture was produced for any member(s) of the SL community. The author
emphasizes that contextual equivalence is not semantic equivalence.
The author introduces the notion of 'translation context' as necessary for
understanding contextual equivalence. The translation context is made up of the
SL context and the TL context. These two contexts have different implications,
possibly resulting in 'contextual interference' and elusiveness of contextual
This final chapter of the book, Chapter 7, is titled 'Symbolic Competence'. This
chapter refers back to the first chapter and aims at explicitly illustrating the
importance for translation of the symbolic subcompetence part of the
intralinguistic competence. The problem, as the author puts it, is that there is
no level of translation known to us whereby the TL community would wish to
identify SL forms with their target language textures adjudged to have the same
forms as source language textures. So, how exactly is symbolic competence
involved in translation? The answer is simply that symbolic competence is
demonstrated at all levels of translation. The author goes on to illustrate
(with examples) this for all the four levels.
The book presents a coherent and clear view of the author's four-level theory of
translation. The major elements of the theory presented in the book can be
* Language is meant essentially for communication and translation is a kind of
communication (Houbert, 1998)
* Language 'construction' and comprehension can be explained in terms of three
Rs (referend, reference and referent)
* Equivalence is the basis of translation (Catford, 1965)
* There are four levels of translation according to the author, as opposed to
the conventional three levels
* The first two levels are SL-oriented, while the last two are TL-oriented
* Communication context can be explained in terms of seven 'variables'
I am not really an expert in translation theory, but as a practitioner of
translation and also as a person working on machine translation, these elements
seem valid to me at the first glance. However, to what extent they are
scientifically valid is something I cannot really say. The book, understandably,
does not quite establish the scientific validity of these elements.
Understandably, because, for one thing, it difficult to devise experiments to
test a translation theory (Adewuni, 2008). But the book does succeed in making a
strong case for the ideas presented.
For example, the idea that language is meant essentially for communication is
something that can be argued only in subjective terms because it is hard to
define what communication means when you include language-thinking, which does
not involve two individuals. I say this because thinking also involves usage of
language which is 'unfinished', or 'incoherent', or 'ungrammatical'. It will be
difficult to draw a line where language begins and non-language-thinking ends.
If you do not include language-thinking, then the idea obviously does not remain
valid. I really do not see the need to restrict language use to 'communication',
however we define it. Ultimately, even in a dialogue between two persons, there
may be an appearance of successful communication, whereas in reality there is
only misunderstanding or partial communication, even if we assume that all that
is said (or written) is 'grammatical' or 'intelligible'. On the other hand, it
is much safer to say that translation is a kind of communication (Houbert, 1998).
Similarly, the idea of equivalence (Catford, 1965) is, as some have argued, not
very suitable to scientific investigation (Leonardi, 2000, Karimi, 2006), unless
you restrict it to lower linguistic levels at which equivalence (or similarity)
can be actually measured quantitatively (Singh & Surana, 2007). This might be
just a limitation of the current state of the science of language, and we might
in the future be able to specify equivalence more accurately (scientifically).
But we are still far from that state, even if it is possible to achieve. Having
said that, I would suggest that the idea of equivalence, when put forward as
just an approximation, can be very useful for practical purposes, e.g. machine
translation (Singh & Husain, 2007). It is from this (computational) point of
view that I find the idea of translation equivalence really exciting.
About the author's contention that there should be four levels instead of three,
the book makes a very strong and convincing case. But I just wonder whether an
equally strong case can be made for, say, five levels instead of four. However,
if we accept that there are indeed four levels, then it is correct to say that
the first two are SL-oriented and the last two are TL-oriented.
The chapter on contextual equivalence is also relevant for computational
linguistics. It can, of course, be useful for translators too. The author's
specification of communication context in terms of seven variables is intuitive,
but not something that can be established very scientifically. In my opinion, it
need not be scientific to be useful for many purposes. I am pointing this out
just because the author has emphasized the scientific nature of the theory
proposed in the book.
On the whole, I found the book an interesting read and I agree with many of the
ideas presented. Even though the author says more than once that real machine
translation is still in the realm of science fiction (and I agree with him), the
theory presented in the book seems more interesting to me from the computational
point of view than as a guide for translators.
There are some minor points that I should mention. The first is that there is
more than necessary repetition in some cases. The second is that some ideas are
too basic for other researchers (though they may be useful for lay readers). And
the ''remedial strategies'' suggested in the chapters about the last two levels
are too sketchy to be very useful, at least in a book which was meant to be
useful for translators.
Even so, this book can be of interest to anyone who is interested in translation
from any point of view, as it has been clearly written, has been illustrated
with good examples, and presents a coherent (author's) view of translation in
Achebe, C. (1958). _Things Fall Apart_. Heinemann.
Adewuni, S. (2008). Linguists and Culture Experts at a Crossroad: Limitations in
Formulating an Experimental Translation Theory. _Translation Journal_ 12 (2).
Arnold, D., Balkan, L., Meijer, S., Humphreys, R. & Sadler, L. (1993). _Machine
Translation: an Introductory Guide_. Blackwells-NCC, London.
Catford, J. C. (1965). _A Linguistic Theory of Translation: An Essay in Applied
Linguistics_. Oxford University Press, Oxford.
Chomsky, N. (1965). _Aspects of the Theory of Syntax_. The MIT Press, Cambridge,
de Saussure, F. (1916). _Cours de Linguistique General_. Payot. (Fromkin, 1999)
Fromkin, V., ed.
Hemingway, E. (1941). _For Whom the Bell Tolls_. Penguin Books.
Houbert, F. (1998). Translation as a Communication Process. _Translation
Journal_ 2 (3).
Hutchins, W. J. & Somers, H. L. (1992). _An Introduction to Machine
Translation_. Academic Press, London.
Karimi, L. (2006). Equivalence in Translation. _Translation Journal_ 10 (1).
Leonardi, V. (2000). Equivalence in Translation: Between Myth and Reality.
_Translation Journal_ 4 (4).
Orwell, G. (1949). _Nineteen Eighteen-Four_. Harcourt Brace.
Sapir, E. (1947). _Selected Writings on Language, Culture and Personality_.
University of California Press.
Savignon, S. (1997). _Communicative Competence: Theory and Classroom Practice_.
McGraw Hill, New York, NY.
Singh, A. K. & Husain, S. (2007). Exploring Translation Similarities for
Building a Better Sentence Aligner. In _Proceedings of the 3rd Indian
International Conference on Artificial Intelligence, Pune, India_.
Singh, A. K. & Surana, H. (2007). Using a Single Framework for Computational
Modeling of Linguistic Similarity for Solving Many NLP Problems. In _Proceedings
of Eurolan Doctoral Consortium, Iasi, Romania_.
Uwajeh, M. K. C. (1994). _Perspectives: Studies in Translatology_
Whorf, B. L. (1956). _Language, Though and Reality. Selected Writings Benjamin
Lee Whorf_. J. B. Carroll edition, Wiley.
ABOUT THE REVIEWER
Anil Kumar Singh is working as a researcher in Natural Language Processing (NLP)
and Computational Linguistics at the Language Technologies Research Centre
(LTRC), IIIT, Hyderabad, India. His research interests are: Modeling and
application of linguistic similarity in the broadest sense, computational
modeling of scripts, computational phonology and morphology, modeling and
processing of temporal information in Natural Languages (NLs), statistical NLP,
machine translation, annotation, corpus linguistics and, last but not the least,
NL engineering. He is the initiator and the main developer of an open source NLP
platform called Sanchay, a collection of tools, APIs and interfaces for NLP,
especially for South Asian languages.