Research questions

This study is the fourth in the series of studies studying sight translation with eye-trackers, starting from Huang (2011) to Chen (2013) and Su (2013). It would follow the methodology of Su (2003) to examine the experienced interpreters’ pause patterns and to triangulate the oral data with the eye-movement data to find eye

fixations during the pauses. With the fixations found, eye movement during the pauses could be studied, and possible cognitive processing inferred. By following Su’s (2013) methodology, the results on the experienced interpreters could be compared with her discoveries on the novices. In doing so, the research aims to answer the following questions.

1) Were experienced interpreters more fluent in sight translating in the sense that they produced fewer pauses than novice interpreters?

2) Were experienced interpreters more fluent in sight translating in the sense that the distribution of juncture pauses and hesitation pauses in their oral output were different from the novice interpreters’?

3) Did experienced interpreter show any difference from novice interpreters in eye movement data such as number of fixations, reading passes and saccade directions during the two types of pauses?

4) What cognitive process could be inferred from data analysis and how was experienced interpreters’ cognitive process different from the novices’?

7

Chapter 2 Literature Review

2.1 Expertise

2.1.1 Definition of expertise

In previous literature, expertise is mostly viewed as excellence (Mieg, 2009). To be more specific, expertise is the ability to constantly produce superior performance (Ericsson and Lehmann, 1996; Ericsson, 2000, 2006; Feltovich, Prietula & Ericsson, 2006). In almost all domains, there are tasks that represent the essence of expertise (Ericsson, 2000). Moreover, through studying superior task performance, features of expert performance in different domains, such as driving, sports, playing chess or music, have been found (Ericsson and Lehmann, 1996; Feltovich et al., 2006; S. E.

Dreyfus, 2004; Sunnari & Hild, 2010). Research has found that experts and their performance in different fields in fact reflect a shared cognitive mechanism of expertise across domains (Feltovich et al., 2006). Compared with novices, experts show distinctive cognitive skills in the way they organize knowledge and respond to task demands (Chi, 2006; Feltovich et al., 2006; Moser-Mercer, 1997). It is hoped that with more findings about traits of expertise in general, the study of expertise could become a science of learning that helps to better students’ performance in different domains (Feltovich et al., 2006; Leighton, 2009).

General cognitive characteristics have been found among experts. Experts have automated some basic components in tasks they perform constantly (Feltovich et al., 2006; Leighton, 2009; Moser-Merser, 2010) so that they not only respond quicker, but could also pay more attention to higher-level skills (Feltovich et al., 2006; Leighton, 2009) and make more use of available knowledge (Feltovich et al., 2006). It has also been found that experts organize knowledge and perceived information in larger

8

chunks and according to the information’s functions in tasks they have to perform.

Moreover, they are capable of restructuring and refining the knowledge and of selecting the most relevant information to adapt to changes in the environment (Feltovich et al., 2006).

Meanwhile, two common myths about expertise have been challenged by researchers: one is the automatic link between professional experience, as a social indicator, and performance-based expertise; the other is the belief that experts were born rather than made (Ericsson, 2000).

Having obtained a certain social status in a field may also be viewed as expertise (Mieg, 2009). However, social indicators such as a person’s reputation, level of training and professional experience do not guarantee a person’s level of expert performance (Ericsson, 2000, 2006). In a well-established profession, where systems of training and evaluation are designed to foster and certify individual expertise (Mieg, 2009), high-performers may have been well-trained and may enjoy high status.

Nevertheless, it is not necessarily true the other way around. Research has found cases in medicine, auditing, mathematics, computer programming and physics, where experienced individuals did not show more accuracy in decision-making (Ericsson &

Lehmann, 1996). Empirical research also found that not all professional translators outputted good-quality translation in think-aloud experiments (Jääskeläinen, 2010).

When people see professionals with similar training or experience perform in various qualities, it is tempting to attribute the differences to talents. Indeed, physical or mental endowments are a common explanation for growth or stagnation of

individual expertise (Ericsson, 2000). It is commonly believed that only individual with certain talents could reach expert levels in a certain field. However, the commonsensical view has been refuted by recent research (Ericsson & Lehmann, 1996; Ericsson, 2000).

9

It would be helpful to distinguish between “absolute expertise” and “relative expertise”, proposed by Chi (2006). Absolute expertise is found in the very few cases of truly exceptional people such as chess masters or high-level athletes, who are

endowed with mental or physical strengths way above most people. Relative expertise, on the other hand, is based on a continuum from novices to experts and on the

assumption that novices can attain expertise and move forward on the continuum.

With training and practice, perceptual, motor and cognitive capacities can be obtained (Ericsson & Lehmann, 1996). What determines expert performance and continuous development of expertise is experience and deliberate practice (Ericsson, 2000, 2006, 2007).

While Ericsson (2000) acknowledged the necessity of extensive experience in reaching expertise in a domain, he also noted that extensive experience didn’t

“invariably lead to expert levels of achievement” (p. 685). It was argued previously that professional experience may only be an indicator of social status rather than performance quality (Ericsson, 2000, 2006). The failure of some professionals to deliver (e.g. Jääskeläinen, 2010) seems to support the view. The discrepancy between the professionals’ experience and expertise raises the question as to the nature of their professional experience.

One possible explanation for the discrepancy may be the distinction between

“routine experts” and “adaptive experts”. Hatano and Inagaki (1986) proposed that people doing something repeatedly would be able to perform procedural skills sufficiently enough to be called “routine experts”. However, they may not have the conceptual knowledge, i.e. systemic understanding of the principles involved in the procedure, to modify or even invent skills to adapt to changes in circumstances.

Expertise involves reflection on the thought processes and methods during task performances (Feltovich et al., 2006). Only when someone has the conceptual

10

knowledge and the flexibility it brings could he or she be called an “adaptive expert”.

Sunnari and Hild (2010) found some professional interpreters could be described as routine experts because their work was based on routines and fixed solutions, which may not promise consistent superior performance. Quality problems were found in some of the professional participants’ SI performance, which showed that these experienced professionals had stagnated at a certain level and fell short of genuine expertise.

2.1.2 Development of expertise

Chi (2006) proposed expertise as “a level of proficiency that novices can achieve”

(p.22). He adapted from Hoffman (1998) a proficiency scale from naïve, novice, apprentice, journeyman, expert to master. Similarly, Hubert L. Dreyfus and Stuart E.

Dreyfus proposed in 1980s a five-stage adult skill acquisition model from novice, advanced beginner, competence, proficiency to expertise (S. E. Dreyfus, 2004). As indicated by the name of the Dreyfus model, adult skills can be acquired.

With training and experience, it is possible to move from one stage to the next on the road to expertise (Chi, 2006; Hoffman, 1998; S. E. Dreyfus, 2004). However, as discussed in the previous section, some people seem to have stopped progressing at various stages before reaching expertise despite cumulative experience (Ericsson, 2000; Jääskeläinen, 2010; Sunnari and Hild, 2010). Individual differences are closely related to the amount of time spent on deliberate practice (Ericsson, Krampe, &

Tesch-Röme, 1993; Ericsson & Lehmann, 1996; Ericsson, 2000, 2006).

Studying expert violinists at a music academy in Berlin, Ericsson et al. (1993) found that while all groups of violinists spent about the same amount of time per week on music-related activities, the best of them spent more time on activities designed to improve performance, or “deliberate practices”.

11

Deliberate practices are practice efforts made to achieve specific goals in performance improvement (Baker, Côté, & Abernethy, 2003; Ericsson et al., 1993;

Leighton, 2009). While the theory of deliberate practice was first developed through research with musicians (Ericsson et al., 1993), it has been applied to numerous fields (Ericsson 2006, 2007) including sports (e.g. Baker et al., 2003) and learning and education (e.g. Leighton, 2009). Gruber, Palonen, & Degner (2008) not only

recognized the importance of deliberate practice to high ability but further studied the social contexts that guided it.

While automaticity is a key feature of expertise as it allows more cognitive resources to be paid to more demanding skills (Feltovich et al., 2006; Leighton, 2009;

Moser-Merser, 2010), when behaviors are automatized, people lose conscious control and mere additional experience do not lead to better performance. Daily activities such as tying shoelaces or getting up from chairs are very good examples (Ericsson, 2006). Deliberate practice involves more demanding tasks that require

problems-solving. Performers stretch their skills, actively adapt their cognitive mechanisms to novel settings and therefore avoid the negative effect of automaticity (Ericsson, 2006; Leighton, 2009). The routine professionals in Jääskeläinen (2010) and Sunnari & Hild (2010) may have considerable experience, but the experience with executing routine tasks, without deliberate efforts to change particular aspects of performance, does not lead to improvement (Ericsson, 2006). To put it simply, they may practice, but they don’t practice deliberately.

2.2 Expertise in interpreting

Like other complex skills, interpreting expertise can be learned through extensive training (Moser-Mercer, 2010). Quantitative findings have shown that interpreting skills are not extension of L1 or L2 skills but are acquired (Sunnari &

12

Hild, 2010). Lambert (2004) described the skill-acquisition process during

interpreting training. At the beginning of training, when all the skill components in the processes need constant attention, students tend to make more mistakes. As the

training moves on, students become more like professionals, for whom many of the mental procedures during interpreting are automatized and thus require less

concentration. Chiang, Kuo, & Chen (2009) compared sight translation performance of beginners and advanced students, who’d had one year of translation and

interpreting training. It was discovered that with training, pause length and frequency in students’ output dropped significantly, so did the influence of language

directionality. Their findings, especially the reduced effect of directionality in the advanced students’ performance, may serve as a support for the role of training on the way to expertise.

Moser-Mercer (2010) proposed the need for 5,000 to 10,000 hours for expertise acquisition in interpreting. Based on the number, she believed that when most trainee interpreters finished their two-to-three-year training program, they were still

journeymen in Hoffman’s model (Hoffmann, 1997) and that they still needed more time on “task in real-life settings for its [professional expertise’s] full development”

(p.265). In other words, extensive post-training experience is important for expertise-acquisition.

In the world of interpreting, experience is a fundamental criterion for expertise.

The AIIC (International Association of Conference Interpreters) only accepts

“competent professional interpreters” as its members. The association does not give tests but requires all applicants to have at least 150 days of work experience to show they have passed the "test" of the workplace (AIIC, n.d.).

The professionals also consider experience important. Tiselius (2010), for

example, interviewed two groups of EU accredited interpreters on their understanding

13

of expertise. While the importance of experience was ranked differently by the two groups in a survey, later discussion revealed experience was deemed important as it would make “both an excellent and average interpreter even better” (P.15).

The majority of the studies on expertise in interpreting equate expertise with professionalism, or professional work experience (Sunnari and Hild, 2010). Dillinger (1994) and Agrifoglio (2004) invited “professional interpreters” to be their subjects representing experienced interpreters. Although it’s unclear if and how they set criteria of experience, the subjects had respectively an average of 3830 hours (Dillinger, 1994) and at least 9 years (Agrifoglio, 2004) of experience in the field.

Chen (2013), on the other hand, adopted the AIIC criterion when looking for experienced interpreters to be her subjects, asking for at least 150 days of work experience.

Many interpreting studies examined and compared performance of interpreters at different stages of training (e.g. Bartłomiejczyk, 2006; Chiang et al., 2009) or with different amount of experience (e.g. Lee, 2011; Chen, 2013; Hu, 2013; Setton and Motta, 2007). Some even involved bilinguals who had no training or experience in interpreting (e.g. Barik 1973, 1975; Christoffels, de Groot, & Kroll, 2006; Köpke &

Nespoulous, 2006). The approach of comparison and contrast was taken in order to find out about behaviors and abilities that changed with expertise development (Liu, 2009). The idea behind the studies was that novices would eventually become experts through training and experience accumulation—in line with the idea of “relative expertise” proposed by Chi (2006). The approach was termed “expert-novice paradigm” by Moser-Mercer, Frauenfelder, Casado, & Künzli (2000).

As discussed before, experts are generally more capable of restructuring knowledge and selecting information most relevant to tasks (Feltovich et al., 2006).

Similarly, expert translators and interpreters’ knowledge seems to show different

14

organization (Moser-Mercer, 1997). Expert translators and interpreters makes better associative connections and domain connections, as well as better semantic

connections so that their interpretation is more contextual-based, in the sense that ideas are better linked and that utterances are made according to speech types and negotiation situations. Novices, on the other hand, treat each utterance as a separate unit and fail to form discourse links (Moser-Mercer et al., 2000). The view is echoed by Liu (2009), who found experts were capable of more flexible semantic processing, e.g. perceiving more important information in and the bigger picture of the source speech, and therefore produced more cohesive interpretation output.

In keeping with the expert-novice paradigm, some researchers proposed further distinctions along the continuum from novice to expertise in interpreting, such as Hoffman (1997). In Hoffman’s development progression model, a “naïve” is someone who knows nothing about a domain. When a person has some introductory exposure to the domain, he or she becomes a “novice”. When a novice has passed the

introductory phase and has had an initiating experience at, for example, consecutive interpreting, he or she has entered the “initiate” level. The next stage is “apprentice”.

An apprentice is a student on a program of translation and interpreting, ranging from 8 months to three years in time. A graduate from one of the programs who have just passed his or her final interpreting exam is a “journey man”, a reliable worker who can perform sufficiently enough in the interpreting booth. With extensive and concentrated experience, a journey man becomes an “expert”, who makes accurate and reliable judgment, shows a lot of skills and economy of effort, and can deal with rare or tough tasks. When an expert is qualified to teach others, he or she reaches the highest level in the model and becomes a “master”.

The finer distinctions on the continuum help to specify the differences between novice and expertise (Hoffman, 1997). The additional skill and experience acquisition

15

at each level illuminate what expertise is, or what experts have while novices don’t.

汝明麗(2010) borrowed the Dreyfus & Dreyfus model (1986) to apply to interpreting teaching and assessment. There are five stages in the model (S. E. Dreyfus, 2004; 汝 明麗,2010): “Novice”, where novice learners with little or no experience rely solely on rules to perform. Novice interpreters start with stock phrases commonly used in conferences and basic rules and skills for interpreting. “Advanced Beginner”, where learners still rely on rules but have started to observe situational aspects. Advanced interpreting learners have better command of the rules and can adjust according to situations although they are not able to judge which rules or situational aspects are the most relevant. “Competence” is the next stage, where learners are aware of goals and can utilize the familiarized rules and previous experience to reach the goals.

Competent interpreters start to see the goal of delivering the message, or the bigger picture, rather than disjointed sentences. They are capable of choosing rules, based on previous experiences, to achieve the goal. “Proficiency”, where the competent

performer are more involved in the task and can make judgments about important aspects of the current situation and make plans to deal with it. Proficient interpreters have a holistic view of the situation, and understand information and judge its importance almost immediately. “Expertise”, where performers are capable of responding to situations and knowing how to achieve goals almost immediately thanks to extensive experience with different situations. Expert interpreters have extensive experience and are flexible with rules so that they can deal with all kinds of situations almost intuitively and effortlessly. 汝明麗(2010) proposed that an

interpreting student must at least reach the competence level to pass the professional exam held at the end of the two-year training. The exam is also regarded as a

minimum criterion for entering the interpreting market. 汝明麗(2010) believed that with the standard established, training institutes could employ teaching resources

16

more efficiently to guide students to meet the required level. On the other hand, judges of the professional exams would have more realistic expectations of the candidates’ skills and performance.

2.3 Sight translation

Interpreting is to translate spoken words from one language into another. It could be broadly divided into two categories: consecutive interpreting (CI) and Simultaneous Interpreting (SI). During consecutive interpreting, the interpreter starts to translate orally after the speaker has finished a sentence or a paragraph in the speech. In simultaneous interpreting, the interpreter listens to the original speech and outputted translation almost at the same time.

Sight translation (ST) is a hybrid of translation and interpretation (Sampaio, 2007). Like in a task of translation, an interpreter/ translator reads a piece of text but instead of putting the translation down in written from, he or she orally outputted the translation, like in an interpreting task.

Despite the debate on whether it is closer to translation or interpreting, sight translation is commonly used in conference, court and community interpreting when interpreters are presented with a piece of text, which may be the speaker’s script or documents that need to be translated orally on the spot to listeners. It is also

recommended and used as a pedagogical tool in early stages of interpreting training as it helps to develop certain skills and cognitive processes needed for consecutive interpreting (CI) and/or simultaneous Interpreting (SI) (Agrifoglio, 2004; Lee, 2012;

Sampaio, 2007; Song 2010). Although used in earlier stages of training, ST itself is no less difficult than SI or CI due to the lack of prosodic aids (such as tone, hesitations and pauses), more complex sentence structures in written texts, and a higher risk of source-language interference (Agrifoglio, 2004).

17

In Gile’s Effort Model (1995), an ST task involved reading and production effort.

Unlike in models for SI and CI, memory effort was missing in Giles’ model. However, although theoretically the interpreter doing the task could always go back to previous sentences or paragraphs she’d read, it is practically not impossible for her to spend too much time on searching and reading old information on paper, otherwise the output flow would be disrupted by long pauses (Sampaio, 2007). Moreover, the interpreter needs to remember what she’d read and said to produce coherent translation and/or grammatically-correct sentences (Agrifoglio, 2004). She may also need to consult information stored in long-term memory, such as background knowledge or terminology (Sampaio, 2007). In short, memory effort is still required.

In addition to memory, previous researchers have also pointed out many skills and strategies shared by ST and SI, such as meaning unit identification, chunking, and anticipation based on linguistic and contextual cues (Song, 2010). Moreover,

compared with ST with prior reading, the time gap between information acquisition and oral output is much shortened in ST without prior reading, so that it resembles SI

compared with ST with prior reading, the time gap between information acquisition and oral output is much shortened in ST without prior reading, so that it resembles SI