Translation theory course (Cak119)
Outline notes for Andrew Chesterman’s lectures
at the MonAKO Programme of Multilingual Communication, University of Helsinki
• What is a theory?
– the Greek origin of the word: ‘a way of looking at something, a point of view; a contemplation’.
– myths as theories: (the Tower of Babel, in Genesis, chapter 11; Eastern myths about rebirth and spiritual development)
– metaphors as theories
– empirical, scientific theories: the importance of testing claims.
• What is translation?
– how wide is your definition? adaptations included?
– bad translation vs. non-translation?
– translation as a prototype concept
• Aims of translation studies
Theoretical aims: to increase understanding... by
– defining central concepts
– describing what translations are like, what translators do
– explaining why translations are like this
– discovering how translations affect readers and cultures
Applied aims: to improve the quality of translations
(and therefore intercultural relations)... by
– developing better training methods
– developing better tools, computer aids, dictionaries, term banks...
– developing machine translation
– developing ways of assessing translation quality
– educating the public, making translators more visible...
• Levels of translation theory
The textlinguistic level (source and target texts, other relevant texts)
The cognitive level (the decision-making process in the translator’s head)
The sociological level (the translation task, its purpose, deadline, the client, the contract...)
The cultural level (ideological factors, power relations, cultural evolution...)
How can we best understand the relation between translations and their source texts?
Basic problem: translations are assumed to be somehow “the same” as the original, but they are obviously different. What does “equivalence” mean?
• Six steps in the debate:
(1) Early Bible translation: equivalence must be understood as sameness. Translation must be possible.
Modern semiotics: translation is always possible (depending on what you mean by translation....)
(2) The problem of untranslatability. Can meaning be separated from form? Absolutely the “same” meaning is (usually / often / sometimes ) impossible. The claim that there are no stable meanings (cf. deconstruction, postmodernism).
(3) Equivalence is better understood as similarity, relevant similarity, not sameness or identity. Two texts can be “similar” in very many ways.
(4) There are several basic types of equivalence: concerning form, meaning, style, desired effect...
(5) Each translation task requires its own hierarchy of equivalence: what are the priorities for this particular translation? Depends on aim and context.
(6) Conclusion: equivalence is produced by the translator, it is claimed. This claim may or may not be accepted by the client, by readers.
How can we best understand the relation between translations and other texts in the target language? This is the relation of naturalness or acceptability, or textual fit. Should translations always be totally fluent?
Arguments for fluency
• Translation has long been part of the tradition of rhetoric in the West.
• Importance of getting the message across, pleasing the reader, making things clear, interpreting
• Early ideas about translating sense for sense
• Translation as adaptation, exegesis
• Bible translation into vernacular languages (Luther, Agricola...)
• Early ideas about the importance of making the translation sound as natural as possible
• The Renaissance and the belles infidèles: translations that were very free indeed: ‘beautiful but not faithful’
• The importance (in modern textlinguistic approaches) of parallel texts (natural, non-translated target-language texts, as far as possible with the same subject and style as the translation). You can study parallel texts in order to see how to increase the naturalness of your translation.
–> all this illustrates the fluency ideal.
There has been criticism of this ideal.
Arguments against fluency
• Fluent translations are often too free, they contain translation errors.
• Fluent translations are unethical, because they deny the Other; they are ethnocentric, they encourage cultural imperialism. Some scholars argue that it would be more ethical to translate in a foreignizing way, not a domesticating one (e.g. Venuti, with respect to literary translation). Foreignizing translations are not fluent, they do not sound entirely natural: that’s the point!
> When could it be acceptable to translate in a non-fluent way?
In what sense are translations different, special? Do they form a code of their own, a third code, distinct both from the source language and from the target language?
• The historical background of this idea, in the Romantic period in Germany
• Language as logos, a creative force, unifying a nation, shaping cognition
• Linguistic relativity (the way you think is affected by the language you speak)
• Translations can stretch and enrich the target language, especially if they are not completely fluent / natural (Schleiermacher)
• Goethe on translation as metamorphosis; retranslation
• Benjamin (1923) on “pure language”, a mystical concept underlying all natural languages; letting the original language shine through in a translation (i.e. not translating fluently)
• Steiner on the “hermeneutic motion” of translation:
–> trust, aggression, incorporation, restitution (= compensation)
Modern corpus studies compare translations (a) with source texts, and (b) with parallel texts. The aim is to generate and test hypotheses about translation universals: generalizations about features that seem special to all translations, making them different from other texts,
• S-universals: differences between translations and their source texts, regardless of language.
Examples: interference, standardization, explicitation
• T-universals: differences between translations and parallel texts in the target language.
Examples: simplification (less lexical variety, lower lexical density), under-representation of TL-specific items
What kinds of translation are there? (Both translators and clients ought to know!)
• Jakobson’s semiotic classification
–> intralingual, interlingual, intersemiotic
• Binary classifications:
– free vs. literal
– covert vs. overt (House)
– semantic vs. communicative (Newmark)
– documentary vs. instrumental (Nord)
• Classification based on text types (Reiss)
–> informative, expressive, operative; multi-medial
• Classification used in the EU: depends on purpose of the translation (Wagner)
– straight (nothing corrected)
– tidied (errors corrected)
– naturalized (adapted to local or international readership)
– reduced (gist translation)
• Towards a more complex typology: specify the variables...
• A typology can be useful in negotiating with clients (and educating them!). They should specify their wishes with respect to different variables, such as translation purpose, content equivalence, formal equivalence, stylistic equivalence, source text revision, level of acceptability required.
• A typology is also useful for scholars wanting to study the general features of particular translation types.
What kinds of textual changes do translators make?
• Strategies are also known as shifts or procedures or techniques.
• Early classifications:
– by Nida: changes of order, omission, structure, addition
– by Vinay and Darbelnet: loan, calque, literal translation, transposition, modulation, total syntagmatic change, adaptation.
• A summary list of some frequent strategies (with Finnish terms added):
Literal translation kirjaimellinen käännös
Transposition sanaluokan muutos
Unit change yksikön muutos (eg. word > phrase)
Structural change rakenteen muutos
Cohesion change koheesion muutos
Rhetorical scheme change retorisen kuvion muutos (alliteration...)
Using a synonym synonyymi
Using an antonym antonyymi
Using a hyponym alakäsite (hyponyymi)
Using a hyperonym yläkäsite (hyperonyymi)
Modulation modulaatio (e.g. concrete > abstract) Rhetorical trope change kielikuvan muutos (metaphor, irony...)
Formality change muodollisuusasteen muutos
Speech act change puheaktin muutos
Transediting toimittaminen, uudelleen muokkaaminen
What might be the best strategies for translating metaphors, allusions, neologisms, names, slang, dialects.... ?
Can machines translate?
• The neopositive belief in laws / rules: language as a well-defined system
• Algorithms; Catford’s conditioned equivalences; translation rules
> under conditions ABC..., item X can be translated as item Y
• Initial optimism of the machine translation project
• Problem of the semantic barrier: in order to interpret sentences correctly, especially ambiguous sentences, we often need to look at the context and appeal to our knowledge of the world.
• The present situation. Examples: SYSTRAN, METEO, Kielikone
• Computer-Aided Translation
– translation memory systems
• By-products of the project: theoretical, technical and practical
How can we best analyse the communication tasks of translators?
Several “functional” theories offer conceptual tools.
• Information theory:
Sender –> Message –> Receiver
• Nida’s sociolinguistic approach:
Analysis –> Transfer –> Restructuring
> structural simplification to kernel forms,
> semantic analysis,
> pragmatic analysis of intended effect, logical structure, stylistic markers)
• Action theory / Skopos theory (Vermeer, Holz-Mänttäri, Nord):
– Social action: writer, client, translator, publisher, reader...
– Skopos (purpose of the translation)
– Specification (client’s instructions)
• The New Rhetoric formula: Who says what, how, when, where, why, to whom, with what effect? (Nord)
• Relevance theory (This is a pragmatic theory, based on work by Sperber and Wilson, about how people communicate: they follow the principle of relevance.)
Relevance: most benefit for least cost
Benefit = effect, importance to receiver
Cost = effort, amount of mental processing by receiver
Effect: depends on receiver’s contextual assumptions
> so, applied to translation (Gutt): translate in such a way as to achieve optimal relevance for the readers; imagine what the readers assumptions are, what they need to be told...
What are the cultural causes and effects of what translators do?
Translating takes place in a cultural context, as part of cultural transfer and evolution.
• A descriptive approach, not prescriptive
–> a model for describing translations (Lambert and van Gorp)
1. Preliminary data (publication data, paratexts etc)
2. Macro-level (major changes; an integral translation?)
3. Micro-level (study of shifts (strategies)
4. Context (relation with other translations, other similar works; reception, reviews...)
• Target-oriented research: starting with the translation itself (Toury)
• Central concepts:
System: a complex of interacting elements, in an environment (Hermans)
Polysystem: a system of systems
Norms: social notions of correctness
– stronger than conventions
– norm-breaking can bring sanctions
– norms vary (time and place)
• Toury’s classification of norms:
– preliminary norms (general translation policy, directness)
– initial norms (source or target-oriented)
– operational norms (textual choices, footnotes, omissions...)
• Definition of translation: a text that conforms to the target-culture’s norms of what translations are supposed to be like, at a given time.
• Ways of explaining translation
– socio-cultural constraints (Lefevere):
> ideology (values, e.g. feminist, postcolonial)
> patronage (who pays?)
> universe of discourse (subject matter; censureship, taboos)
> poetics (literary conventions)
> language-pair differences (contrastive analysis)
– laws (generalizations, general hypotheses)
> Toury’s laws of interference and growing standardization
What goes on in the translator’s head?
• Another source of possible explanations
• Influence of Game theory (about how people make decisions)
• How to get inside the Black Box?
– Vary input and check changes in output
– Measure involvement
– Study drafts (interim solutions analysis)
– Computer studies of time distribution, keystrokes
– Think-aloud protocols (TAPs)
–> models of the problem-solving process
• Some key results
> attention units
> non-linear processing
> routine vs. non-routine processes
> influence of self-image
> influence of emotional state
> differences between professionals and amateurs
What is a good translation?
• Error analysis
– definition of error; different criteria
– error types
– scale of gravity
• Retrospective assessment (comparing with the source text)
– to measure equivalence
– e.g. House
• Prospective assessment (measuring the effect)
– e.g. Nida: tests of comprehension etc.
– analysis of effects:
> ± intended
> de dicto (form) or de re (content)
> “same” as source text or not
– effect on whom?
– NB: prescriptive statements are hypotheses of effect.
• Introspective assessment (TAPs)
• Lateral assessment (comparing with non-translated parallel texts)
– checking translation against expectancy norms
– relevance of corpus research
• International standards: quality control of the process
What values influence translators’ decisions?
• Contractual theories (rights and duties)
vs. utilitarian theories (depends on results)
• Personal vs. professional ethics
• Descriptive (what are the norms and values?)
vs. prescriptive (what should they be?)
• Macro-ethics (social level: the world)
vs. micro-ethics (textual level: the word)
Example of ethical problem: errors in the source text (Wagner)
– do nothing
– literal translation + [sic]
– correct covertly
– correct overtly (add translator’s note)
– correct fully (also in source)
– check with writer or client!
Models of translation ethics
• Ethics of representation
– representing the source, the Other; loyalty to author
• Ethics of service
– meeting client’s needs
• Ethics of communication
– co-operating with the Other
– loyalty to the translator’s profession
• Norm-based ethics
– meeting expectations
> Norms and associated values (AC)
• Expectancy norm: meet expectations of readers
• Relation norm: get appropriate relation with source text
• Communication norm: optimize communication
• Accountability norm: be loyal to all parties
• Problems with each model
• Hyvä kääntämistapa (Finnish Translators’ Association)
• The Translator’s Charter
Do we have a theory? Depends what you mean by a theory...
In the philosophy of science, a scientific theory
– describes, explains and predicts...
– ...via laws (high-level generalizations)
– develops by generating and testing hypotheses
– may use metaphors and models in order to create a useful conceptual framework
– may be more or less formalized
• Different kinds of laws:
– universal (absolute) vs. probabilistic (more or less probable)
– theoretical (based on internal logic, like in maths) vs. empirical (based on observable evidence)
• Different kinds of explanation
– causal (previous conditions) vs. teleological (function, purpose)
– generalizing / subsumptive (explaining via appeal to a higher-level general law)
vs. reductive (explaining via relating to lower-level phenomena)
– “causes” (external) vs. “reasons” (internal, mental)
– contextualizing (showing relation with broader context)
• Aristotle’s causes:
material (target language, source text)
efficient (mind and body of the translator)
• In translation studies, no single theory covers everything.
We have lots of theory, but not A Theory.
> Description? Lots of useful concepts, e.g. strategies...
> Explanation? Analyses of many kinds of probabilistic causes...
> Prediction? E.g. knowledge about training methods that work...
> Laws? A few proposals...
> Hypotheses? A few... They need much more testing!
> Metaphors? Lots!
> Models? Several, e.g. our four-level model (see lecture 1).
> Formalized? No. Perhaps not necessary.
Back to Andrew Chesterman – homepage.