The Göpferich Model: The Development of Translation Competence

The Göpferich model of translation competence is concerned less with what translation ability is made of than with how it changes over time. That difference of emphasis is the feature that sets the model apart from other competence frameworks.

Contents

Published by Susanne Göpferich in 2009, the model was developed as the theoretical framework for a longitudinal research project called TransComp. By following translation students over a period of years, the project set out to see whether competence really does develop, and in what order.

Below we look at the components of the model, at the concepts peculiar to it such as self-concept and routine activation, at the tendencies the research brought to light, and at what all of this means for the party buying translation.

What does the model try to explain?

Most competence models take a photograph: which components are present in a competent translator? The Göpferich model tries to shoot a film. Its question is this: how does the relationship between those components change over the course of training?

There is a practical reason for asking it. Translator training programmes teach their material in a particular order and assume that the order works. Testing that assumption requires development to become measurable.

The second reason is individual difference. Two students take the same classes but progress at different speeds. The model looks for the source of that gap in the balance between the components: in one student language knowledge is ahead, in another it is the habit of research.

The third reason is the problem of measurement. Scoring the quality of a translation says nothing about how the process ran. The model proposes looking at process data instead: where does the translator pause, which resource do they turn to, when do they change a decision?

This approach forms part of a wider effort to move translator training away from examinations and towards observation. What is measured is not the product but the route to the product.

The components of the model

The model defines translation competence as a set of components that feed one another. The table below sets out the components and what they amount to in everyday terms.

Component	What it means	Sign of development
Communicative competence	Advanced production and comprehension in at least two languages	Fewer sentences in the target language that look like the source language
Domain competence	Knowledge of the subject of the text and knowing where to obtain that knowledge	Narrower research, faster arrival at the right source
Tools and research competence	Dictionaries, corpora, translation memory and search techniques	The form of the query matching the problem
Activating translation routines	Reaching ready-made solutions for structures that come up often	Less hesitation over formulaic structures
Psychomotor competence	Speed of writing, reading and interface use	Less time spent on the keyboard and with the tool
Strategic competence	Planning, monitoring and correcting the process	Planned experiment in difficult passages rather than random searching

Alongside these components the model has two external elements: the translator’s self-concept and their psycho-physical disposition. These determine how the components will be used, and they cannot be taught directly.

The model also takes in working conditions: the time available, the resources that can be used and the client’s instructions. The same competence produces a different result under different conditions, which is why it is misleading to measure competence by looking at a single translation.

Self-concept: how does the translator see themselves?

Self-concept is the picture a translator holds of their own role. Someone who sees themselves as a faithful conduit for the source text and someone who sees themselves as the spokesperson for the target reader will take different decisions on the same sentence.

In the model this concept is not decoration; it is a filter that sets the direction of decisions. A translator with a weak self-concept hands responsibility back to the source and takes refuge in word-for-word translation. That way they believe they have made no mistake.

Another dimension of self-concept is a sense of professional responsibility. A translator who accepts that they carry the consequences of the text they deliver behaves differently when it comes to confidentiality, accuracy and delivery on time.

Psycho-physical disposition, meanwhile, covers attributes such as attention, memory, patience and tolerance of stress. These cannot be taught directly, but they can be supported by working conditions. An interrupted workflow and poor interfaces drain this resource quickly (Ehrensberger-Dow and O’Brien 2015).

The implication for training is this: self-concept cannot be made the subject of a lesson, but it is shaped by feedback. The self-concept of a student who is asked to defend a decision develops differently from that of a student who only receives corrections.

When do routines help, and when are they a trap?

The competence of activating translation routines is the most visible product of experience. For structures that occur often in a given language pair, ready-made solutions form in the translator’s mind and come into play without deliberation.

These routines save time. The set phrases of contracts, the warning sentences of user manuals and the connectives of academic writing do not have to be solved afresh every time.

A routine becomes a trap, however, as soon as it ignores context. The habitual solution engages quickly and the translator does not notice that the text is saying something else. This is why experienced translators make a different kind of mistake from novices.

The same risk grows when technology is added to it. A high-percentage match coming out of a translation memory appears to confirm the routine already in the mind. Bowker (2005) describes this excess of confidence as blind faith in the translation memory.

For this reason the model places strategic competence above the routines. Strategic competence is the supervisory layer that decides when a routine should be stopped.

What did the longitudinal research show?

The TransComp project followed translation students at set intervals throughout their training, using the same methods each time, and compared the resulting data with data from professional translators. The methods used were think-aloud recordings, screen recordings and keystroke logging.

The value of a longitudinal design is this: in a one-off measurement, the student who translates well cannot be told apart from the student who is developing. Measure the same person again and again and you see the direction of travel.

One of the study’s prominent findings is that development is not linear. Students appear to go backwards in certain periods, because they begin to see problems they had not noticed before and the process slows down. That slowing is itself a sign of development.

The second finding is that the components do not develop at the same rate. Tools and research competence develops relatively quickly, while strategic competence and self-concept settle more slowly. This explains why technology classes produce results in a short time yet are not enough on their own.

The third finding is that individual differences are large. The gap between students on the same programme is more marked than the gap between programme averages. That increases the importance of individual feedback.

The limits of the model

The first limit of the model follows from the nature of the data collection method. Thinking aloud slows the translator’s process down and creates an unnatural working environment. The process that is measured differs somewhat from the real one.

The second limit is the number of participants. Longitudinal studies are necessarily run with small groups; it is hard to find large numbers of people who can be followed for years. This limits how far the findings can be generalised.

The third limit is the language pair. The model was developed on data centred on German and English. In languages with a different typology, such as Turkish, the weight of the components may change.

The fourth limit is the pace of technology. The model was developed before today’s AI-assisted tools became widespread. The definition of tools and research competence has broadened since then.

These limits notwithstanding, the model’s contribution endures: it teaches us to stop treating competence as a list and to think of it as a process. That view changes both the design of training and the criteria for recruitment; it makes it possible to assess seniority not by a number of years but by the component in which it has accumulated.

The Türkiye context

Assessment on translation programmes in Türkiye is generally product-focused: the translation that is handed in is given a mark. The process-focused observation the Göpferich model proposes can be applied in small groups with instruments such as screen recording and a process log, but in crowded classes the method is expensive.

One point specific to Turkish concerns routines. Turkish is an agglutinative language, and long sentences in the source language have to be broken up in Turkish. Because that splitting decision recurs so often it quickly becomes a routine, and once it does there is a risk of approaching every text with the same template.

The practice of sworn translation, for its part, adds a local dimension to the discussion of self-concept. A translator who has sworn an oath before a notary is personally responsible for the accuracy of the text. That responsibility strengthens the faithful-conduit self-concept, and with official documents this is usually the outcome that is wanted.

In marketing and corporate communications texts, by contrast, the same self-concept can be limiting. The same translator has to adopt a different role in the two kinds of work, and that switch does not happen by itself; it is a conscious decision. This article is for general information and is not legal advice.

What does this mean for you?

The model gives substance to what experience means. A translator’s number of years tells you nothing on its own; the real question is which component the experience has accumulated in. Someone who has translated contracts for twenty years may not have translated a single marketing text in those same years.

When choosing a supplier, you might try this sequence:

Define the type of your text: does it convey information, does it persuade, or does it produce legal effects?
Ask what work the candidate has done in that type over the past year; do not settle for general seniority.
In a test translation, ask for the reasoning behind the decisions; self-concept becomes visible here.
Establish at the outset whether you will be asked questions when something is unclear.
On long-running work, prefer to stay with the same team; routines are only of use to you once they have been built on your own texts.

These points let you match the right seniority to the right job. We have summarised the way we work by field on our services page; if you would like to work out the right profile for your own text together with us, you can use the quotation form.

Frequently asked questions

How does the Göpferich model differ from the PACTE model?

Both divide competence into components; the difference lies in the emphasis. The Göpferich model puts development and the movement between components at the centre, and positions self-concept and psycho-physical disposition as separate elements.

What does the competence of activating translation routines mean?

It is the ability to reach quickly for the ready-made translation solutions that are used often in a given language pair. These routines save time, but they turn into a source of error when they do not fit the context.

What does longitudinal research mean?

It is the kind of research in which the same people are followed at set intervals over a period of years. Unlike one-off measurements, it can show the direction and the speed of development.

Is this model of any use to me when I buy translation?

It is. The model shows the areas in which experience genuinely makes a difference, so that you can judge more accurately which jobs a translator’s seniority actually matters for.

References

Göpferich, S. (2009). Towards a Model of Translation Competence and Its Acquisition. In: Behind the Mind. Copenhagen: Samfundslitteratur, pp. 11-37.
Bowker, L. (2005). Productivity vs Quality? Localisation Focus, 4(1), pp. 13-20.
Ehrensberger-Dow, M. and O’Brien, S. (2015). Ergonomics of the Translation Workplace. Translation Spaces, 4(1), pp. 98-118.
Pietrzak, P. and Kornacki, M. (2021). Using CAT Tools in Freelance Translation. London: Routledge Focus, Chapter 5.