تويت
Think-Aloud Protocols; A Method to Investigate the Translation Process
As it is not possible to directly observe the human mind at work, a number of attempts have been made to access the translator’s mind indirectly. One such attempt has been to ask the translators themselves to reveal their mental processes in real time while performing a translation task. This article is aimed to explore the methodological aspects of think-aloud protocols as a method to investigate the complex process of translation.
Key words
Think-aloud protocols, translation processes, data collection, verbalization.
Introduction
Human cognitive processes – and particularly the mental process of translation – can be investigated in different ways such as observing reaction to specific stimuli, analyzing the errors and the results of a task performance, etc. Over the last three decades, think-aloud protocols (TAPs) have become a widely-used method to investigate the complex process of translation. This method allows data collection about the translator’s thoughts at the same time he verbalizes them.
The method of thinking aloud consists in organizing an experiment in which subjects are asked to carry out a task and to verbalize their thoughts while performing. The task performance is recorded on audio- or video-tape. The resulting recordings are then transcribed (think-aloud protocols, or TAPs) and subjected to analysis.
The theoretical framework for TAP experiments is provided mainly by the work of Ericsson and Simon (1984). According to their model, humans keep information in different memory stores, characterized by different access and storage capabilities: short-term memory (STM) presents easy access but extremely limited storage space, whereas long-term memory (LTM) presents larger storage space but more difficult access (Bernardini, p. 1999).
The STM is also called working memory (WM) which is the primary site of the procedural memory. LTM, by contrast, serves as the vessel for the declarative memory (M. A. Schmidt, 2005). As far as the translation process is concerned, it is important to consider the function and capacity of the STM because translating relies on declarative knowledge as well as procedural knowledge. “This distinction is crucial because the cognitive processes, as well as information that is not currently being processed, cannot be reported but must be inferred by the analyst on the basis of the verbalizations” (Bernardini, 1999, p. 2).
Implications of Ericson and Simon’s model
First of all, according to Ericsson and Simon’s model, only concurrent verbalization of thoughts could efficiently reflect the mental states of a subject performing a relatively long task; a task which takes longer than ten seconds to complete, according to Ericsson and Simon (Bernardini, 1999). It is important to notice that a cognitive process takes longer time when the subject thinks aloud. “This means that people are able to slow down the normal process to synchronize it with verbalization” (Someren, Barnard and Sandberg, 1994, p. 33).
When the subject has completed the task, some parts of the information may move on to LTM, leaving behind some retrieval cues in STM. It has been found that in such cases the later verbalization may be difficult and often incomplete (Ericsson and Simon, 1993). Moreover, under these circumstances, it can be extremely problematic to know that whether the subject is generating their thought processes in a new and different way or retrieving them exactly from LTM.
Secondly, to make sure that the subject actually reports his mental states without distorting them, it is important that he does not feel he is taking part in a social interaction. Although conversation is obviously a natural situation in human life, it may involve some reworking thoughts to adjust them to socially established norms; this process might change the actual information to some extent (Bernardini, 1999). Emotional and motivational factors can also produce a cognitive process different from the one that would take place without thinking aloud. The interaction between the subject and the researcher should therefore be avoided or at least reduced to a minimum. But there is one other cause for concern; if the subject keeps silent for a long time, the verbalization will become useless, because significant parts of the cognitive process in STM may not be investigated. To avoid this, the researcher is allowed to repeat to the subject to think aloud, although with a short and non-interfering speech; Ericsson and Simon propose to use the phrase “keep talking” (Krahmer and Ummelen, 2004).
Thirdly, “practice and experience may affect the amount of processing carried out in STM, so that fewer mental states will be available for verbalization to subjects experienced in a task” (Bernardini, 1999, p. 2). This process, known as ‘automation’ refers to the fact that “as particular processesbecome highly practiced, they become more and more fully automated” (Ericsson and Simon, 1984, p. 15) and do not require active processing in working memory, i.e. they are performed at an unconscious level and become less accessible for verbalization. “To give a simple example, a novice driver has to focus all of his attention to driving; after most of the process involved in driving has become automatized, it is possible to engage in a conversation while driving” (Jääskeläinen 1999, p. 59). Although these kinds of processes are more difficult to access, it is possible for the researcher to instruct the subjects in order to bring them back to the conscious attention.
There exist several other obstacles on access to the process, for example, a heavy cognitive load during a task performance. Due to STM’s storage limitations, subjects tend to stop verbalizing if they have to pay attention to many things at the same time. “In some cases, processing uses all the available capacity and none is left for producing verbalizations” (Jääskeläinen, 1999, p. 59). For example, if reasoning takes place in a smooth way, verbalizing the contents in working memory is easy and doesn’t use too much memory capacity. However, if the information is complicated, verbalization will take time and space in working memory because it becomes a cognitive process itself. Consequently, the report of the original process may be incomplete and the process itself can even be disturbed (Someren, Barnard and Sandberg, 1994).
Finally, Ericsson and Simon take into account the effects of personality and personal background over the data collected through TAPs. Individual differences in knowledge and ability to verbalize thoughts can heavily bias the data collected through TAPs. The problem here is the object of study and not the methodology used. Individual differences exist, and research could not conceal them. However, it seems advisable to try to limit the influence ofindividual differences and to take them into account when analyzing the data, in order to obtain more reliable and generalizable data (Bernardini, 1999).
All these limitations imply that verbalizations may lose a fraction of the total amount of mental activities occurring at any moment in time. However, this would not reduce the model’s importance in research. The application of think-aloud protocolscontinues to be valid, because this experimental model, used for years in cognitive psychology, would efficiently allow us to observe and analyze the students’ translation processes. Moreover, these reports can be well completed with other kinds of data, such as questionnaires, product analysis, etc. (Jääskeläinen, 1999).
TAPs in translation studies
The analysis of think-aloud protocols (TAPs) in Translation Studies began in Europe in the late 1980s. Scholars felt the necessity to develop empirical and inductive methods in order to take the place of deductive models of the translation process presented until then, which usually described what ideally happens or rather what should happen in translating. There were researchers like Krings and Lörscher in Germany, Dechert and Sandrock in Britain, Jääskeläinen and Tirkkonen-Condit in Finland, who began to ask what actually happens when people translate.
This new trend can be explained in some parts by developments in other disciplines such as psychology, which had renewed its interest in the study of the mental process with some other appropriate and legitimate methods of research. This change had an impact on psycholinguistic researches, including research on second language learning and translation studies (Jääskeläinen, 1999).
“There has always been a kind of empirical research, like translation criticism and error analysis, but this was product- and not process-oriented” (Kussmaul and Tirkkonen-Condit, 1995, p. 177). By comparing the target text with the source text or looking at errors of the translated text, one could only speculate about what has occurred in the translator’s mind during translation. What was needed was a way to discover what actually happens, “to get a glimpse into the ‘black box’, as it were” (Kussmaul and Tirkkonen-Condit, 1995, p. 178).
Viewing translation mainly as a problem-solving activity, some scholars proposed that it should be possible to study it by means of TAPs, and set up experiments to test this hypothesis. However, the different interests and backgrounds of the researchers involved have resulted in a large variety of independent approaches (Bernardini, 1999).
It should be mentioned that this kind of analyses increases our potential for describing and explaining the translation processes. Moreover, they have at least two pedagogical purposes. (1) The different strategies observed in the TAPs may serve as good models for successful translating (Lörscher, 1992; Jääskeläinen, 1993; Krings, 1988; Kussmaul, 1993). (2) If translation students are used as subjects, TAPs may be used to find out where they have problems. The data collected can then form a basis for translation pedagogy (Krings, 1988; Kussmaul, 1989a+b, 1994). It might be argued that teachers of translation, from years of experience, already know which strategies they should recommend to students. But they are likely to draw wrong conclusions from their students’ translations. Teachers may, for instance, have the impression that students have problems with text-comprehension while, when talking to them, they find that students actually have problems in expressing what they had understood. TAPs can help teachers to see matters more clearly (Kussmaul and Tirkkonen-Condit, 1995, p. 178-9).
In reality, one of the first areas to apply verbal reports regarding the study of language use was research on foreign language (FL) learning. Other areas were research on writing processes and on FL reading processes. Some of the researches on L2 learning/acquisition, for example, have used translation tasks to elicit data on students’ text processing strategies (Gerloff, 1986) or on the organization of cognitive planning in a translation task. Consequently, these studies may offer interesting insights and hypothesis for translation-oriented researches.
It is important to know what are the general aims and the experimental details of different TAP studies, in order to assess their findings and relate them to other kinds of research. The first essential distinction is to see whether the emphasis of a research is on translation studies or psycholinguistics. In other words, it should be clear that whether the aim is to understand the nature of translating or whether translation is used as an experimental task to collect data on the nature of language processing (Jääskeläinen, 1999).
First studies on foreign language learners
The first studies conducted by Sandrock (1982) and Krings (1986) already show the advantages and the limitations of this method of data collection and set the standards for the design of other similar studies. Kings’ study of the translation processes of eight advanced students of French shows the immense richness of data that can be obtained by TAPs.
However, from the point of view of Translation Studies, the research has a drawback. The participants in this study are not involved in translation as a professional activity; they are, in fact, foreign language students who have translated their tasks the same way they usually translate an ordinary assignment in a language class.
According to Krings and Gerloff, the results of their experiments reveal that inexperienced translators, in fact, employ more local strategies, which are concerned only with the fragment they are working on, without considering the text as a whole. Moreover, they don’t relate to their own world-knowledge. More experienced translators, in contrast, use more global strategies, which relate the problem to their world-knowledge, to the text as a whole and to its overall theme.
Lörscher (1991) investigated the translation process in foreign language learners. He used first- and second-year students of English at the university he was working at as the subjects of his experiment (they were not even advanced learners). Assuming that oral translation would provide richer material than written translation, he instructed his participants to translate a written text orally and recorded their spoken translations, including all concurrent verbalizations. With this research, Lörscher claims to investigate the translation process itself, even if he recognizes that his model does not resemble a “real mediating situation”, as he calls it because:
In all of these experiments, the subjects were mostly foreign language learners, rather than students of translation, which has received a fair amount of criticism, because their findings can hardly account for professional translators’ performance. However, they provided important information about translating by foreign language learners and laid the methodological basis for later TAP studies, which investigated the distinctions among non-professional, semi-professional (translation students) and professional translation (Königs, 1987; Kussmaul, 1998; Jonasson, 1998).
Further studies: different aims and hypotheses
Following these pioneers, a number of other translation researchers have used TAP to elicit data in their studies. These studies have often different settings and involve different subject populations (translation students, professional translators, teachers of translation, laypersons, bilinguals, a combination of these categories); different language pairs (depending on where the research has been carried out); different types of task and experimental conditions (translating a written text orally, producing a written translation, translating alone, translating in pairs or in small groups; translating with or without access to reference materials, limited or unlimited available time); different text-types (political texts, newspaper editorials, tourist brochures, government documents); different categories of analysis (identification of translation problems and problem-solving strategies, focus on conscious attention, role of affective factors on translation).
Pöntinen and Romanov (1989) organized a TAP experiment with two subjects: the first one was a teacher of translation and free-lance translator; the second one was a subject specialist. They were both about the same age with a high level of education. The data collected showed some interesting differences between the two subjects’ decision criteria: the translator relied more on textual knowledge than the subject specialist (Jääskeläinen, 1999). The purpose of the first TAP experiments at Savolinna was to identify differences between professional vs. non-professional translation. The terms ‘professional’ and ‘non-professional’ referred, misleadingly, to fifth-year and first-year students of translation respectively. However, it has been frequently pointed out that the differences between first-year and fifth-year students’ translation processes may not depend on different levels of translation competence alone, but on other factors such as differences in their world knowledge.
Some of the most recent TAP studies on the translation process wanted to discover what kind of linguistic and extra linguistic factors influence the production of “good” translations. Jensen (2000) and Jääskeläinen looked at the influence of routine vs. non-routine tasks on task performance and investigate the differences between professionals and laymen. Künzli (2003), furthermore, explored the impact of emotional and affective states on subjects’ performance. Several of these studies are aimed to investigate the differences between categories of translators such as professionals, advanced students in translation training programs and language students with respect to their translational behavior (Norberg, 2003).
Some researchers in Denmark proposed methods of recording the writing process during translation to develop and complement data collected by means of TAP. Their studies point towards a possible design combining different analyzing methods that could be able to elicit and evaluate data, telling us more about the complex structures that govern the translation process (Schmidt, 2005).
It should be mentioned that due to the difficult and time-consuming methods of data collection and analysis involved in TAP research, the numbers of subjects have remained relatively small (ranging from one to 48) in these studies and investigators have been extremely careful in generalizing on the basis of the data collected.
Conclusion
To conclude this article, we should state that the presence of different researches in this issue has the advantage to focus on different aspects of different kinds of translation processes. This increases our understanding of the complex mechanisms underlying translation. However, the differences in the kinds of data collected, analyses carried out, and the overall goals of research have made it more difficult to compare the methods employed in previous experiments from all points of view.
References
Bernardini, S. (1999). Using think-aloud protocols to investigate the translation process: methodological aspects. Bologna: University of Bologna.
Ericson, K. A., & Simon, H. A. (1984/1993). Protocol Analysis. Verbal Repoprts as Data. Cambridge, MA and London: Massachusetts Institute of Technology.
Gerloff, P. (1986). Second language learners’ reports on the interpretive process: talk-aloud protocols of translation.In J. House & S. Blum-Kulka (Eds.), Interlingual and Intercultural Communication (pp. 243-262). Tübingen: Gunter Narr.
Jääskeläinen, R. (1993). Investigating translation strategies. In Y. Gambier & J. Tommola (eds.), SSOTT IV. Translation and Knowledge (pp. 99-120). Turku: University of Turku.
Jääskeläinen, R. (1999). Tapping the process: an explorative study of the cognitive and affective factors involved in translating. Joensuu: University of Joensuu.
Jonasson, K. (1998). Degree of Text Awareness in Professional vs. Non-professional translators. In A. Beylard-Ozeroff, J. Králová & B. Moser-Mercer (eds.), Translators’ Strategies and Creativity: Selected Papers from the 9<SUP>th </SUP>International Conference on Translation and Interpreting (pp. 189-200). Amsterdam: John Benjamins.
Königs, F. G. (1987). Was beim Übersetzen passiert: Theoretische Aspekte, empirische Befunde und praktische Konsequenzen. Die Neueren Sprachen.
Krahmer, E., & Ummelen, N. (2004). Thinking About Thinking Aloud. A comparison of two verbal protocols for usability testing, Netherlands: Tilburg University.
Krings, H. P. (1986). Translation problems and translation strategies of advanced German learners of French (L2).In J. House & S. Blum-Kulka (Eds.), Interlingual and Intercultural Communication (pp. 263-276). Tubingen: Gunter Narr.
Krings, H. P. (1988). Blick in die ‘Black Box’ – Eine Fallstudie zum Übersetzungsprozeß bei Berufsübersetzern. In R. Anrtz (ed.), Textlinguistik und Fachsprache Akten des Internationalen übersetzungswissenschaftlichen AILA-Symposions (pp. 393-412).Hildesheim: Olms.
Krings H. P. (1988). Thesen zu einer empirischen Übersetzungswissenschaft.In J. Holz-Mänttäri (Ed.), Translationstheorie – Grundlagen und Standorte (pp. 58-71). Tampere: Tampereen Yliopisto
Kussmaul, P. (1989a). Toward an Empirical Investigation of the Translation Process: Translating a Passage from S. I. Hayakawa: Symbol, Status and Personality. In R.Von Bardeleben (369-380).
Kussmaul, P. (1989b). Interferenzen im Übersetzungsprozess – Diagnose und Therapie. In H. Schmidt (19-28).
Kussmaul, P. (1993). Empirische Grundlagen einer Übersetzungsdidaktik: Kreativität im Übersetzungsprozeß. In J. Holz-Mänttäri & C. Nord (Eds.), Traducere navem (275- 286). Geburtstag. Tampere: Universitätsbibliothek, S.
Kussmaul, P. (1998). Die Erforschung von Übersetzungsprozessen: Resultate und Desiserate. München: Lebende Sprachen.
Kussmaul, P., & Tirkkonen-Condit, S. (1995). Think-Aloud Protocol analysis in translation studies. TTR – Traduction Terminologie Rédaction 8(1), 177-191.
Lörscher, W. (1991). Translation Performance, Translation Process, and Translation Strategies. A Psycholinguistic Investigation. Tübingen: Günter Narr Verlag.
Lörscher, W. (1992). Process-Oriented Research into Translation and Implications for Translation Teaching. Neubrandenburg: University of Greifswald.
Norberg (2003) Übersetzen mit doppeltem skopos. Eine empirische Prozess und produktstudie. Uppsala: Uppsala University Library.
Pontinen, T., & Romanov, T. (1989). Professional vs. non-professional translator: a think- aloud protocol study. Joensuu, University of Joensuu, Savonlinna School of Translation Studies.
Sandrock, U. (1982). Thinking-aloud protocols (TAPs). Ein Instrument zur Dekomposition des Komplexen Prozesses “Übersetzen”. Stockholm: Kassel Svenska Akademiens ordlista.
Schmidt, M. A. (2005). How do you do it anyway? A longitudinal study of three translator students translating from Russian into Swedish. Stockholm: Acta Universitatis Stockholmiensis.
Someren, M. W., Barnard, Y. F., & Sandberg, J. A. C. (1994). The think aloud method. A practical guide to modeling cognitive processes. London: Academic Press.
By Iman Poostdoozan,
Islamic Azad University,
Fars Science and Research Branch, Shiraz, Iran
Iman_motarjem@yahoo com
========
Abstract Islamic Azad University,
Fars Science and Research Branch, Shiraz, Iran
Iman_motarjem@yahoo com
========
As it is not possible to directly observe the human mind at work, a number of attempts have been made to access the translator’s mind indirectly. One such attempt has been to ask the translators themselves to reveal their mental processes in real time while performing a translation task. This article is aimed to explore the methodological aspects of think-aloud protocols as a method to investigate the complex process of translation. Key words
Think-aloud protocols, translation processes, data collection, verbalization.
Introduction
Human cognitive processes – and particularly the mental process of translation – can be investigated in different ways such as observing reaction to specific stimuli, analyzing the errors and the results of a task performance, etc. Over the last three decades, think-aloud protocols (TAPs) have become a widely-used method to investigate the complex process of translation. This method allows data collection about the translator’s thoughts at the same time he verbalizes them.
The method of thinking aloud consists in organizing an experiment in which subjects are asked to carry out a task and to verbalize their thoughts while performing. The task performance is recorded on audio- or video-tape. The resulting recordings are then transcribed (think-aloud protocols, or TAPs) and subjected to analysis.
It is important to note, however, that “thinking aloud” as a method of eliciting data is not the same as “thinking aloud” in the everyday sense; it entails more than sitting people down next to tape-recorder and asking them to talk (Jääskeläinen, 1999, p. 9).
The Ericson and Simon’s modelThe theoretical framework for TAP experiments is provided mainly by the work of Ericsson and Simon (1984). According to their model, humans keep information in different memory stores, characterized by different access and storage capabilities: short-term memory (STM) presents easy access but extremely limited storage space, whereas long-term memory (LTM) presents larger storage space but more difficult access (Bernardini, p. 1999).
Information input will first be heeded by the STM and when its capacity and storage time is exhausted, the information is transferred to the LTM. A certain loss both prior and during this transfer is assumed, but it does not seem to be a substantial loss (M. A. Schmidt, 2005, p. 27).
It is argued that only information which is present in STM can be directly accessed and reported; LTM contains information which has left consciousness, but can be retrieved later back to STM for more processing.The STM is also called working memory (WM) which is the primary site of the procedural memory. LTM, by contrast, serves as the vessel for the declarative memory (M. A. Schmidt, 2005). As far as the translation process is concerned, it is important to consider the function and capacity of the STM because translating relies on declarative knowledge as well as procedural knowledge. “This distinction is crucial because the cognitive processes, as well as information that is not currently being processed, cannot be reported but must be inferred by the analyst on the basis of the verbalizations” (Bernardini, 1999, p. 2).
Implications of Ericson and Simon’s model
First of all, according to Ericsson and Simon’s model, only concurrent verbalization of thoughts could efficiently reflect the mental states of a subject performing a relatively long task; a task which takes longer than ten seconds to complete, according to Ericsson and Simon (Bernardini, 1999). It is important to notice that a cognitive process takes longer time when the subject thinks aloud. “This means that people are able to slow down the normal process to synchronize it with verbalization” (Someren, Barnard and Sandberg, 1994, p. 33).
When the subject has completed the task, some parts of the information may move on to LTM, leaving behind some retrieval cues in STM. It has been found that in such cases the later verbalization may be difficult and often incomplete (Ericsson and Simon, 1993). Moreover, under these circumstances, it can be extremely problematic to know that whether the subject is generating their thought processes in a new and different way or retrieving them exactly from LTM.
Secondly, to make sure that the subject actually reports his mental states without distorting them, it is important that he does not feel he is taking part in a social interaction. Although conversation is obviously a natural situation in human life, it may involve some reworking thoughts to adjust them to socially established norms; this process might change the actual information to some extent (Bernardini, 1999). Emotional and motivational factors can also produce a cognitive process different from the one that would take place without thinking aloud. The interaction between the subject and the researcher should therefore be avoided or at least reduced to a minimum. But there is one other cause for concern; if the subject keeps silent for a long time, the verbalization will become useless, because significant parts of the cognitive process in STM may not be investigated. To avoid this, the researcher is allowed to repeat to the subject to think aloud, although with a short and non-interfering speech; Ericsson and Simon propose to use the phrase “keep talking” (Krahmer and Ummelen, 2004).
Thirdly, “practice and experience may affect the amount of processing carried out in STM, so that fewer mental states will be available for verbalization to subjects experienced in a task” (Bernardini, 1999, p. 2). This process, known as ‘automation’ refers to the fact that “as particular processesbecome highly practiced, they become more and more fully automated” (Ericsson and Simon, 1984, p. 15) and do not require active processing in working memory, i.e. they are performed at an unconscious level and become less accessible for verbalization. “To give a simple example, a novice driver has to focus all of his attention to driving; after most of the process involved in driving has become automatized, it is possible to engage in a conversation while driving” (Jääskeläinen 1999, p. 59). Although these kinds of processes are more difficult to access, it is possible for the researcher to instruct the subjects in order to bring them back to the conscious attention.
There exist several other obstacles on access to the process, for example, a heavy cognitive load during a task performance. Due to STM’s storage limitations, subjects tend to stop verbalizing if they have to pay attention to many things at the same time. “In some cases, processing uses all the available capacity and none is left for producing verbalizations” (Jääskeläinen, 1999, p. 59). For example, if reasoning takes place in a smooth way, verbalizing the contents in working memory is easy and doesn’t use too much memory capacity. However, if the information is complicated, verbalization will take time and space in working memory because it becomes a cognitive process itself. Consequently, the report of the original process may be incomplete and the process itself can even be disturbed (Someren, Barnard and Sandberg, 1994).
Finally, Ericsson and Simon take into account the effects of personality and personal background over the data collected through TAPs. Individual differences in knowledge and ability to verbalize thoughts can heavily bias the data collected through TAPs. The problem here is the object of study and not the methodology used. Individual differences exist, and research could not conceal them. However, it seems advisable to try to limit the influence ofindividual differences and to take them into account when analyzing the data, in order to obtain more reliable and generalizable data (Bernardini, 1999).
All these limitations imply that verbalizations may lose a fraction of the total amount of mental activities occurring at any moment in time. However, this would not reduce the model’s importance in research. The application of think-aloud protocolscontinues to be valid, because this experimental model, used for years in cognitive psychology, would efficiently allow us to observe and analyze the students’ translation processes. Moreover, these reports can be well completed with other kinds of data, such as questionnaires, product analysis, etc. (Jääskeläinen, 1999).
TAPs in translation studies
The analysis of think-aloud protocols (TAPs) in Translation Studies began in Europe in the late 1980s. Scholars felt the necessity to develop empirical and inductive methods in order to take the place of deductive models of the translation process presented until then, which usually described what ideally happens or rather what should happen in translating. There were researchers like Krings and Lörscher in Germany, Dechert and Sandrock in Britain, Jääskeläinen and Tirkkonen-Condit in Finland, who began to ask what actually happens when people translate.
This new trend can be explained in some parts by developments in other disciplines such as psychology, which had renewed its interest in the study of the mental process with some other appropriate and legitimate methods of research. This change had an impact on psycholinguistic researches, including research on second language learning and translation studies (Jääskeläinen, 1999).
“There has always been a kind of empirical research, like translation criticism and error analysis, but this was product- and not process-oriented” (Kussmaul and Tirkkonen-Condit, 1995, p. 177). By comparing the target text with the source text or looking at errors of the translated text, one could only speculate about what has occurred in the translator’s mind during translation. What was needed was a way to discover what actually happens, “to get a glimpse into the ‘black box’, as it were” (Kussmaul and Tirkkonen-Condit, 1995, p. 178).
Viewing translation mainly as a problem-solving activity, some scholars proposed that it should be possible to study it by means of TAPs, and set up experiments to test this hypothesis. However, the different interests and backgrounds of the researchers involved have resulted in a large variety of independent approaches (Bernardini, 1999).
It should be mentioned that this kind of analyses increases our potential for describing and explaining the translation processes. Moreover, they have at least two pedagogical purposes. (1) The different strategies observed in the TAPs may serve as good models for successful translating (Lörscher, 1992; Jääskeläinen, 1993; Krings, 1988; Kussmaul, 1993). (2) If translation students are used as subjects, TAPs may be used to find out where they have problems. The data collected can then form a basis for translation pedagogy (Krings, 1988; Kussmaul, 1989a+b, 1994). It might be argued that teachers of translation, from years of experience, already know which strategies they should recommend to students. But they are likely to draw wrong conclusions from their students’ translations. Teachers may, for instance, have the impression that students have problems with text-comprehension while, when talking to them, they find that students actually have problems in expressing what they had understood. TAPs can help teachers to see matters more clearly (Kussmaul and Tirkkonen-Condit, 1995, p. 178-9).
In reality, one of the first areas to apply verbal reports regarding the study of language use was research on foreign language (FL) learning. Other areas were research on writing processes and on FL reading processes. Some of the researches on L2 learning/acquisition, for example, have used translation tasks to elicit data on students’ text processing strategies (Gerloff, 1986) or on the organization of cognitive planning in a translation task. Consequently, these studies may offer interesting insights and hypothesis for translation-oriented researches.
It is important to know what are the general aims and the experimental details of different TAP studies, in order to assess their findings and relate them to other kinds of research. The first essential distinction is to see whether the emphasis of a research is on translation studies or psycholinguistics. In other words, it should be clear that whether the aim is to understand the nature of translating or whether translation is used as an experimental task to collect data on the nature of language processing (Jääskeläinen, 1999).
First studies on foreign language learners
The first studies conducted by Sandrock (1982) and Krings (1986) already show the advantages and the limitations of this method of data collection and set the standards for the design of other similar studies. Kings’ study of the translation processes of eight advanced students of French shows the immense richness of data that can be obtained by TAPs.
However, from the point of view of Translation Studies, the research has a drawback. The participants in this study are not involved in translation as a professional activity; they are, in fact, foreign language students who have translated their tasks the same way they usually translate an ordinary assignment in a language class.
Thus, the objective of Kings’ study was translating in a pedagogical context or didactic translation, which is a rather different task than translation as a professional activity. Nevertheless, the study provided a number of research questions and categories to apply to analysis as well as a highly fruitful way to use TAP in the study of the translation process (Schmidt, 2005, p. 22).
Gerloff published her study in 1986. She investigated and compared the translation processes of three different subject populations: four college students of French, four bilingual speakers English/French without any experience of translation, and four professional translators, normally translating from French into English. The study, however, focuses on translation in one direction i.e. from L2 (French) to L1 (English). In this study, Gerloff uses the same coding and classifying categories as Krings. The most important finding is that more experienced translators (experience here is defined in the context of translating being an inherent ability in bilinguals), such as both the professionals and the bilinguals in her sample, do not necessarily translate more easily or faster than the less experienced translators (here defined as the foreign language students). She also concludes, from other indicators, that experienced translators are more aware of the difficulty of the problems they find and of their possible solutions; furthermore, they set higher standards for their performance than novices. According to Krings and Gerloff, the results of their experiments reveal that inexperienced translators, in fact, employ more local strategies, which are concerned only with the fragment they are working on, without considering the text as a whole. Moreover, they don’t relate to their own world-knowledge. More experienced translators, in contrast, use more global strategies, which relate the problem to their world-knowledge, to the text as a whole and to its overall theme.
Lörscher (1991) investigated the translation process in foreign language learners. He used first- and second-year students of English at the university he was working at as the subjects of his experiment (they were not even advanced learners). Assuming that oral translation would provide richer material than written translation, he instructed his participants to translate a written text orally and recorded their spoken translations, including all concurrent verbalizations. With this research, Lörscher claims to investigate the translation process itself, even if he recognizes that his model does not resemble a “real mediating situation”, as he calls it because:
[...] it is still unknown whether translation processes in real mediating situations are different – in detail or in principle – from translation processes in artificial mediating situations (Lörscher, 1991, p. 4).
Despite some inconsistencies between the design and the aim of his study, Lörscher could partly develop a refined model for analyzing TAP, providing a useful tool for further researches. In all of these experiments, the subjects were mostly foreign language learners, rather than students of translation, which has received a fair amount of criticism, because their findings can hardly account for professional translators’ performance. However, they provided important information about translating by foreign language learners and laid the methodological basis for later TAP studies, which investigated the distinctions among non-professional, semi-professional (translation students) and professional translation (Königs, 1987; Kussmaul, 1998; Jonasson, 1998).
Further studies: different aims and hypotheses
Following these pioneers, a number of other translation researchers have used TAP to elicit data in their studies. These studies have often different settings and involve different subject populations (translation students, professional translators, teachers of translation, laypersons, bilinguals, a combination of these categories); different language pairs (depending on where the research has been carried out); different types of task and experimental conditions (translating a written text orally, producing a written translation, translating alone, translating in pairs or in small groups; translating with or without access to reference materials, limited or unlimited available time); different text-types (political texts, newspaper editorials, tourist brochures, government documents); different categories of analysis (identification of translation problems and problem-solving strategies, focus on conscious attention, role of affective factors on translation).
Pöntinen and Romanov (1989) organized a TAP experiment with two subjects: the first one was a teacher of translation and free-lance translator; the second one was a subject specialist. They were both about the same age with a high level of education. The data collected showed some interesting differences between the two subjects’ decision criteria: the translator relied more on textual knowledge than the subject specialist (Jääskeläinen, 1999). The purpose of the first TAP experiments at Savolinna was to identify differences between professional vs. non-professional translation. The terms ‘professional’ and ‘non-professional’ referred, misleadingly, to fifth-year and first-year students of translation respectively. However, it has been frequently pointed out that the differences between first-year and fifth-year students’ translation processes may not depend on different levels of translation competence alone, but on other factors such as differences in their world knowledge.
Some of the most recent TAP studies on the translation process wanted to discover what kind of linguistic and extra linguistic factors influence the production of “good” translations. Jensen (2000) and Jääskeläinen looked at the influence of routine vs. non-routine tasks on task performance and investigate the differences between professionals and laymen. Künzli (2003), furthermore, explored the impact of emotional and affective states on subjects’ performance. Several of these studies are aimed to investigate the differences between categories of translators such as professionals, advanced students in translation training programs and language students with respect to their translational behavior (Norberg, 2003).
Some researchers in Denmark proposed methods of recording the writing process during translation to develop and complement data collected by means of TAP. Their studies point towards a possible design combining different analyzing methods that could be able to elicit and evaluate data, telling us more about the complex structures that govern the translation process (Schmidt, 2005).
It should be mentioned that due to the difficult and time-consuming methods of data collection and analysis involved in TAP research, the numbers of subjects have remained relatively small (ranging from one to 48) in these studies and investigators have been extremely careful in generalizing on the basis of the data collected.
Conclusion
To conclude this article, we should state that the presence of different researches in this issue has the advantage to focus on different aspects of different kinds of translation processes. This increases our understanding of the complex mechanisms underlying translation. However, the differences in the kinds of data collected, analyses carried out, and the overall goals of research have made it more difficult to compare the methods employed in previous experiments from all points of view.
References
Bernardini, S. (1999). Using think-aloud protocols to investigate the translation process: methodological aspects. Bologna: University of Bologna.
Ericson, K. A., & Simon, H. A. (1984/1993). Protocol Analysis. Verbal Repoprts as Data. Cambridge, MA and London: Massachusetts Institute of Technology.
Gerloff, P. (1986). Second language learners’ reports on the interpretive process: talk-aloud protocols of translation.In J. House & S. Blum-Kulka (Eds.), Interlingual and Intercultural Communication (pp. 243-262). Tübingen: Gunter Narr.
Jääskeläinen, R. (1993). Investigating translation strategies. In Y. Gambier & J. Tommola (eds.), SSOTT IV. Translation and Knowledge (pp. 99-120). Turku: University of Turku.
Jääskeläinen, R. (1999). Tapping the process: an explorative study of the cognitive and affective factors involved in translating. Joensuu: University of Joensuu.
Jonasson, K. (1998). Degree of Text Awareness in Professional vs. Non-professional translators. In A. Beylard-Ozeroff, J. Králová & B. Moser-Mercer (eds.), Translators’ Strategies and Creativity: Selected Papers from the 9<SUP>th </SUP>International Conference on Translation and Interpreting (pp. 189-200). Amsterdam: John Benjamins.
Königs, F. G. (1987). Was beim Übersetzen passiert: Theoretische Aspekte, empirische Befunde und praktische Konsequenzen. Die Neueren Sprachen.
Krahmer, E., & Ummelen, N. (2004). Thinking About Thinking Aloud. A comparison of two verbal protocols for usability testing, Netherlands: Tilburg University.
Krings, H. P. (1986). Translation problems and translation strategies of advanced German learners of French (L2).In J. House & S. Blum-Kulka (Eds.), Interlingual and Intercultural Communication (pp. 263-276). Tubingen: Gunter Narr.
Krings, H. P. (1988). Blick in die ‘Black Box’ – Eine Fallstudie zum Übersetzungsprozeß bei Berufsübersetzern. In R. Anrtz (ed.), Textlinguistik und Fachsprache Akten des Internationalen übersetzungswissenschaftlichen AILA-Symposions (pp. 393-412).Hildesheim: Olms.
Krings H. P. (1988). Thesen zu einer empirischen Übersetzungswissenschaft.In J. Holz-Mänttäri (Ed.), Translationstheorie – Grundlagen und Standorte (pp. 58-71). Tampere: Tampereen Yliopisto
Kussmaul, P. (1989a). Toward an Empirical Investigation of the Translation Process: Translating a Passage from S. I. Hayakawa: Symbol, Status and Personality. In R.Von Bardeleben (369-380).
Kussmaul, P. (1989b). Interferenzen im Übersetzungsprozess – Diagnose und Therapie. In H. Schmidt (19-28).
Kussmaul, P. (1993). Empirische Grundlagen einer Übersetzungsdidaktik: Kreativität im Übersetzungsprozeß. In J. Holz-Mänttäri & C. Nord (Eds.), Traducere navem (275- 286). Geburtstag. Tampere: Universitätsbibliothek, S.
Kussmaul, P. (1998). Die Erforschung von Übersetzungsprozessen: Resultate und Desiserate. München: Lebende Sprachen.
Kussmaul, P., & Tirkkonen-Condit, S. (1995). Think-Aloud Protocol analysis in translation studies. TTR – Traduction Terminologie Rédaction 8(1), 177-191.
Lörscher, W. (1991). Translation Performance, Translation Process, and Translation Strategies. A Psycholinguistic Investigation. Tübingen: Günter Narr Verlag.
Lörscher, W. (1992). Process-Oriented Research into Translation and Implications for Translation Teaching. Neubrandenburg: University of Greifswald.
Norberg (2003) Übersetzen mit doppeltem skopos. Eine empirische Prozess und produktstudie. Uppsala: Uppsala University Library.
Pontinen, T., & Romanov, T. (1989). Professional vs. non-professional translator: a think- aloud protocol study. Joensuu, University of Joensuu, Savonlinna School of Translation Studies.
Sandrock, U. (1982). Thinking-aloud protocols (TAPs). Ein Instrument zur Dekomposition des Komplexen Prozesses “Übersetzen”. Stockholm: Kassel Svenska Akademiens ordlista.
Schmidt, M. A. (2005). How do you do it anyway? A longitudinal study of three translator students translating from Russian into Swedish. Stockholm: Acta Universitatis Stockholmiensis.
Someren, M. W., Barnard, Y. F., & Sandberg, J. A. C. (1994). The think aloud method. A practical guide to modeling cognitive processes. London: Academic Press.