Accuracy: Omissions in consecutive versus simultaneous interpreting

Evy Cox
Graduate of KU Leuven


Heidi Salaets
KU Leuven

All views or conclusions are those of the authors of the articles and not necessarily those of the editorial staff or the publisher.
Creative Commons License - Attribution, Non-Commercial, No-Derivatives

[av_font_icon icon=’ue84d’ font=’entypo-fontello’ style=” caption=” link=’manually,’ linktarget=” size=’80px’ position=’right’ animation=’deactivated’ color=” av_uid=’av-1wqvxp’ admin_preview_bg=”][/av_font_icon]
Download PDF


It is sometimes assumed that consecutive interpreting is more accurate than simultaneous interpreting due to note-taking (Andres, 2002; Gile, 2001a; Matyssek, 1989). So far, little comparative research has been conducted to confirm this belief. Omission in interpreting is a well-known phenomenon, which has already been studied from various perspectives. Omissions are a complex matter, especially because a certain degree of subjectivity can never completely be avoided, as will be explained more broadly in this research article. Furthermore, most research on omissions in interpreting consists of small-scale studies. The study described here set out to compare non-strategic omissions in both consecutive and simultaneous interpreting, with a focus on cognitive load. Nine trainee interpreters’ consecutive and simultaneous target text renditions were transcribed and subsequently analyzed. The analysis adopted Kopczynski’s approach (1980) which allowed for greater insight into the contrast between these interpreting modes. It also allowed the researcher to explore how the interpreter mentally approached the source text on a linguistic level. The analysis not only provided a way of categorizing omissions but also allowed for additional information to be extracted from it. The findings of this small study suggest that students’ consecutively interpreted renditions were 15% more likely to show omissions than their simultaneously interpreted renditions. In this study, the memory factor may have played a larger role than the cognitive load factor. An attempt has been made to explain the findings and draw some implications for interpreter education.
Keywords: omission, accuracy, interpreting mode, simultaneous, consecutive, cognitive load, effort models

1. Introduction

Omissions and quality in consecutive interpreting (CI) and simultaneous interpreting (SI) are frequently researched, interrelated concepts. Some scholars have suggested that consecutive interpreting is more accurate than simultaneous interpreting because the former mode allows interpreters to take notes to help them remember what was said (Andres, 2002; Gile, 2001a; Matyssek, 1989). However, there has not been much comparative research to support this assumption. To the best of my knowledge, not many studies have focused on omissions in both CI and SI. This study will contribute to filling this gap by presenting the results of a study on omissions in consecutive and simultaneous interpreting with the aim of assessing accuracy and quality in interpreting.
Previous research has shown an interest in the difficult notion of ‘interpreting quality,’ omissions, note-taking and other related topics. Nevertheless, the current research niche in comparative studies of these interpreting modes came to exist because each of those concepts has generally been studied separately. Gile and Korpal argue that more comparative research on consecutive and simultaneous interpreting should be done (Gile, 2009; Korpal, 2012). In this respect, Korpal stresses the value of comparative research in the field of omissions: “The comparison of the two interpreting modes with respect to omission would make it possible to formulate some general conclusions about the nature of the two modes.” (Korpal, 2012, p. 111)
Korpal demonstrates the practical relevance of this type of research. If we learn more about the nature of both interpreting modes and how they correspond or differ, this may in turn have implications for interpretation and for their inclusion in interpreter training programs.
The aim of the current study was to compare the occurrence or the strategic use of omissions in both interpreting modes. To this end, the author conducted an experiment with nine interpreter trainees. The study was given an extra dimension by adopting Kopczynski’s (1980) conceptual research model. I will start by providing a theoretical framework for the current research. Cognitive load, a central concept in interpreting, will serve as a starting point.

2. Literature review

2.1 Cognitive load in consecutive and simultaneous interpreting

Cognitive load in interpreting is considered an important factor influencing the overall accuracy of an interpretation (Gile, 2009), and cognitive load varies between consecutive interpreting (CI) and simultaneous interpreting (SI). In this paper, omissions are considered as a factor affecting the accuracy of delivery. Cognitive load might impact the frequency of omissions and so provide an explanation for differences in accuracy between the two interpreting modes. The internal and external factors that affect cognitive load, and, therefore, the accuracy of the interpretation, will be discussed in section 2.3. First, I present a short overview of the cognitive load active in both interpreting modes, based on Gile’s Effort Models (2009).

2.1.1 Effort Models

Gile’s Effort Models (2009) have become an influential research model with regard to the various cognitive effects in both consecutive and simultaneous interpreting (Seeber, 2015). Gile intended his Effort Models to present a pedagogical framework capable of clarifying interpreting errors by looking at mental processing capacity. This implies that there is still insufficient empirical evidence to support the underpinnings of the Effort Models (Gile, 1997, 2008, 2009; Seeber, 2015). Nevertheless, some studies appear to confirm some of the hypotheses upon which the Effort Models were built (Gile, 1999). More research is needed in order to make general assertions.
First of all, Gile makes a distinction between automated and non-automated mental processes. Automated mental processes require little or no cognitive effort (e.g., correct pronunciation in one’s mother tongue). The non-automated mental processes, on the other hand, can exhaust available mental processing capacity. These non-automated processes stem from the content of the source speech, often unknown to the interpreter (Gile, 2009). Gile points out that the Effort Models provide a simplified representation, since interpreting in reality is not that clear-cut: mental processes can be gradually more or less automated (cf. Gile, 1995a).
Gile (2009) represents the non-automated mental processes in formulas (see Appendix 3). The production effort does not only comprise the reproduction in the target language, but also self-corrections made by the interpreter while s/he is listening carefully to his/her own interpretation.
Gile assumes that there is a cognitive load for each individual effort. Added together, they result in a total combined cognitive load. In other words, more cognitive processes imply a higher overall cognitive load. This places higher demands on the limited mental processing capacity of the interpreter. If the available mental processing capacity is exceeded, a decrease in interpretation quality occurs. That is why the total available mental processing capacity must be greater than or equal to the ‘required’ mental processing capacity. Similarly, the available processing capacity for each individual cognitive effort must be greater than or equal to the required processing capacity for that effort (Gile, 2009).
Gile suggests that the overall cognitive load is higher in simultaneous interpreting than in consecutive interpreting. Gile (2001b; 2009) describes two separate stages of consecutive interpreting, namely listening and production (Gile, 2001b; Gile, 2009). However, in simultaneous interpreting, both efforts take place at the same time, whereas in consecutive interpreting they take place successively. Consequently, the interpreter continuously has to keep up with the original speaker in simultaneous interpretation. For consecutive interpreting, this is only the case in the listening stage. This way, cognitive saturation is less likely to occur in the production stage (Gile, 2009). The memory effort also differs between the two modes: in consecutive interpreting, the interpreter has to keep the source text information in mind, processing it, for a much longer time than in simultaneous interpreting. The note-taking effort, however, aims to partially relieve the interpreter’s memory and free up processing capacity (Albl-Mikasa, 2007; Gile, 2009).

2.2 Defining ‘accuracy’

This article will discuss omissions in consecutive versus simultaneous interpreting within the broader framework of accuracy. Since ‘accuracy’ is a much-disputed topic in the field of interpreting studies, it is all the more important to clearly define the concept. There does exist consensus about the position of accuracy as part of the broader concept of ‘quality’. I will start by discussing quality, followed by accuracy as a subcomponent of quality. I will then outline the definition of accuracy I will apply in this paper.

2.2.1 Quality

The overarching concept of ‘quality’ is afforded more discussion than its hyponym ‘accuracy.’ The subjectivity of quality (Kalina, 2005) and the variety of possible perspectives on the matter make a unanimous definition nearly impossible (cf. Kahane, 2000). Grbic attributes the complexity of empirical research into quality to the combination of three axiomatic features: the variability of performance; the relativity of judgments depending on the evaluators’ perspective(s); and the relativity of the evaluation process with reference to certain criteria (Grbic, 2015).
In interpreting research, variability of performance takes the individual differences between interpreters into account. Interpreters find themselves in settings in which they have to make individual cognitive, linguistic, and ethical choices, amongst others. These choices shape ‘quality’ and thus hinder a uniform definition of the concept in relation to interpreting (Grbic, 2015).
The relativity of judgments depending on the evaluators’ perspective(s) emphasizes the broadness of the concept of quality. It refers to the multitude of possible perspectives on quality. This equally stands in the way of a uniform definition (Grbic, 2015). When evaluating an interpretation, the evaluator has to choose from a wide range of assessment criteria. An overall assessment comprising all possible criteria would be hardly feasible. For example, assessment criteria could include commitment, professionalism, personality, and interpretation techniques, to mention just a few (Kutz, 1997).
The relativity of the evaluation process with reference to certain criteria involves the users of interpreting services, who all have a different view on what ‘quality’ should look like. This depends on the type of user, the social context, and personal opinion (Gile, 1995a). Over the years, much research on user expectations has been carried out (Bühler, 1986; Kurz, 2001, 2003; Pöchhacker, 2015, 2016). Researchers have found that, while there is consensus among users of interpreting services as to the quality criteria to be assessed (e.g., pleasant voice, thorough preparation, pleasant appearance, fluency, reliability, sense consistency, logical cohesion, completeness, correct grammar etc.), the importance individual users attach to those criteria differs (Gile, 2009).
As a result, some scholars have sought to approach ‘quality’ more generally. This way, they could capture the objective parameters that underlie the subjective choices made when assessing interpreting quality (Garzone, 2002; Kahane, 2000; Pöchhacker, 1993, 1994). However, research in this field so far remains mostly theoretical.

2.2.2 Accuracy

In this paper, ‘accuracy’ will be defined in accordance with Tiselius’ dichotomy (2015). According to Tiselius, accuracy is measured in two complementary manners: first, as the sum of the various elements that build the interpretation (e.g., omissions, additions, fluency); and secondly, holistically, approaching interpretation as an intrinsic whole. The comprehensiveness and transparency of this model allow the existing and more confined approaches to be included.
With regard to Tiselius’ first measurement method –the interpretation of elements – interpreters should strive for accuracy of interpretation. This means few if any omissions or additions, optimal fluency and so forth. It is an approach that underlines correspondence with the source, although it is not a novel one (cf. Gile, 1995b; Kopczynski, 1994; Pöchhacker, 2002; Shlesinger, 1997).
Various authors have designed classifications of those individual interpretation components regarding source-target correspondence. However, Barik’s classification (1971), despite being one of the oldest, probably remains the most well-known. He compiled an elaborate classification of omissions, additions and substitutions/errors that was initially innovative. Nevertheless, due to the subjectivity involved, it was not found to be viable (Barik, 1971).
Given that the focus of this article is on omissions in interpreting, it may be useful to take a closer look at more specific classifications of omissions. Napier (2004, 2015) formulated one such classification and distinguished different types of omissions depending on the interpreter’s awareness level: conscious strategic omissions, conscious intentional omissions, conscious unintentional omissions, conscious receptive omissions and unconscious omissions. Napier gives an example of a classification that captures the underlying objective parameters leading to omissions in interpreting (see 2.2.1): consciousness and intention. However, putting Napier’s model into practice may present some difficulties, simply because it is difficult to determine the interpreter’s level of awareness and intention.
Due to these limitations in other models, the classification that will be used as a guideline for analyzing the data in this study is Kopczynski’s (1980). Kopczynski divides omissions into different types according to the linguistic phrases that are influenced by the omissions, as explained below. Kopczynski makes a distinction between the omission of the core of the phrase (for instance a noun or a verb) or its modifier(s) (e.g., an adjective modifying a noun or an adverb modifying a verb) (Kopczynski, 1980).
Kopczynski’s (1980) analysis has resulted in two categories (noun phrase and verb phrase) being most commonly affected, and some categories (conjunctions, omissions within a conventional syntagm, and other omissions) being less affected. Omissions of complete and unfinished sentences are treated as a separate category. Within the different types of phrases, there are four overarching reasons as to why Kopczynski’s approach serves as a guideline for this study. Firstly, Kopczynski’s proposal goes beyond simply determining what kind of omission occurs most and which interpreting mode is more prone to omissions; his approach of subdividing omissions according to the phrases that are being influenced by them enables a view into how the interpreter mentally analyzes the source text on a linguistic level while aiming at a reproduction in the target language (Kopczynski, 1980). Secondly, his classification succeeds in providing objective parameters. He defines omissions in accordance with the linguistic phrases they affect. Thirdly, it is a methodologically feasible approach. And lastly, Kopczynski uses a descriptive model by examining the omissions made and creating a corresponding classification.
In his own research, Kopczynski’s found that the participating interpreters tried to preserve the main information of the source text by omitting modifiers first. Nonetheless, he stated that more research was needed to test the hypothesis (Kopczynski, 1980).
Tiselius’ second and complementary measurement method – approaching the interpretation as an intrinsic whole (Tiselius, 2015) – is the aspect that makes this approach so comprehensive. It allows the existing and more confined approaches to be included. Although Tiselius does not elaborate on this aspect of her accuracy dichotomy, I posit that Tiselius’ ‘interpretation as an intrinsic whole’ can be compared to what Shlesinger calls the intratextual level of quality, concerning the acoustic, linguistic and logical aspects of the interpretation product per se (Shlesinger, 1997).

2.2.3 Accuracy regarding the present research

As mentioned above, this study is restricted to the component ‘omission’ within accuracy. The research data are analyzed in accordance with Kopczynski’s approach to further test his premise (see 2.2.2). An additional distinction between functional and non-functional omissions is made, since not every omission implies a decrease in accuracy. Sometimes omissions are necessary to generate an accurate and acceptable interpretation (Barghout, Rosendo & García, 2015; Garzone, 2002; Korpal, 2012; Pio, 2003; Tiselius, 2015). Some authors distinguish functional and non-functional omissions in line with consciousness and intention of the interpreter (Barghout, Rosendo & García, 2015; Barik, 1975; Jones, 1998). The feature of consciousness or intention will not be examined in this study, however, since it is very difficult to identify (see 2.2.2). Therefore, all omissions that provoke a loss of significant information, and thus a decrease in accuracy, are considered non-functional. All omissions that boost the interpretation quality are considered functional. An example of a functional omission, collected from the analyzed data for this study:
Source text for simultaneous interpreting: […] we need to find a way to cut the production of single-use and disposable plastics […]
Interpreting student (Dutch): […] we need to find a way to cut the production of disposable plastics […]
It is not necessary to repeat ‘single-use plastics’ and ‘disposable plastics’ twice, since they mean the same thing. In Dutch, this would even be considered a stylistic error. Therefore, the interpreting student made a functional omission by omitting one of the two.

2.3 Influence of cognitive load on accuracy

This section discusses which factors influence cognitive load (see 2.1) and thus accuracy (see 2.2), and to what extent the impact on accuracy differs for each interpreting mode. Gile (2009) defines both occasional reasons and chronic reasons. The term ‘occasional reasons’ refers to triggers that hinder successful interpretation by leading to cognitive saturation, while the term ‘chronic reasons’ refers to inadequate cognitive skills or lack of knowledge. Only occasional reasons (Gile, 2009) were taken into account. These triggers act on cognitive load, which varies according to the interpreting mode, consecutive or simultaneous. Elements which would be defined by Gile as chronic reasons were not considered.

2.3.1 Factors influencing cognitive load in CI and SI

A lot of common factors that affect interpreting quality in general have already been mentioned in earlier research (Albl-Mikasa, 2007; Altman, 1994; Barghout, Rosendo & García, 2015; Barik, 1971, 1973, 1975; De Groot, 1997; Gerver, 1975; Gile, 1983, 1995a, 1997, 2009; Jones, 1998; Kopczynski, 1980; Korpal, 2012; Napier, 2004; Pio, 2003; Pöchhacker, 2016; Pradas Macías, 2003; Shlesinger, in Pöchhacker, 2016). However, research into triggers that affect consecutive and simultaneous interpreting differently or triggers that are unique to either interpreting mode is rather limited. So far, only a handful of studies have associated certain factors that are detrimental to cognitive load when working in a particular interpreting mode. For simultaneous interpreting those factors are ear-voice span, words without a direct translation, multi-word names, incomplete or ambiguous segments at the beginning of sentences, language pairs with major linguistic and/or cultural differences, pauses at ungrammatical points, large grammatical differences between languages, and poor technological equipment for simultaneous interpreting (Issa, 2018; Barghout, Rosendo & García, 2015; Barik, 1975; Gile, 2001a; Jones, 1998). By way of the term pauses at ungrammatical locations, Barik (1975) refers to the speaker pausing when his/her utterance is still incomplete or not complete enough for the interpreter to assign a certain meaning to it. In some cases, the interpreter is forced to start interpreting with insufficient information. For consecutive interpreting, those factors are the span between the speaker and the interpreter taking notes, enumerations and the manual nature of note-taking (Gile, 2001a; Gile, 2009). The role of long-term memory might also be added.

2.3.2 Impact on accuracy in consecutive vs. simultaneous interpreting

Less is known about how the above-mentioned factors affect accuracy in both interpreting modes. It is sometimes assumed that consecutive interpreting is more accurate than simultaneous interpreting due to note-taking (Andres, 2002; Gile, 2001a; Matyssek, 1989). So far, little comparative research has been conducted to confirm this belief. The existing comparative research points to the opposite conclusion. Gile (2001a) asked three professional interpreters to assess a number of consecutive and simultaneous interpretations, resulting in all three considering the simultaneous interpretations to be more accurate. Kopczynski (1980) conducted comparative research for the language pair English-Polish and found that in consecutive interpreting about 20% more omissions occurred at phrase level compared to simultaneous interpreting. Those results indicate that more research is required on the assumption that consecutive interpreting is more accurate than simultaneous interpreting (Gile, 2001a).

3. Methodology

3.1 Participants

I set up this study with nine participants who were Postgraduate Conference Interpreting students at the KU Leuven Antwerp Campus and the University of Antwerp. All students had Dutch as their native language and three students specified English as their B-language (see Appendix 4). Two students had not taken part in the Master in Interpreting program at the above-mentioned universities although most students come to the Postgraduate program for Conference Interpreting after completing a Master’s program. Students who do not are designated “external” participants. I conducted this study as a fellow student and in the context of a Master in Interpreting thesis. With regard to interpreting experience, I took the position of those students who had previously completed the Master in Interpreting as a point of departure. The students had had approximately eighteen months’ experience with consecutive interpreting and three to four months’ experience with simultaneous interpreting. Table 1 shows the number of practice hours reflecting those periods of time. It was difficult to assess the overall training level of the two external students.
Table 1: Students’ experience with CI and SI respectively

Students of the KU Leuven Antwerp Campus Students of the University of Antwerp
Hours of practice for consecutive interpreting 150 250
Hours of practice for simultaneous interpreting 60 80

3.2 Data

The students were asked to interpret two different English source texts into Dutch, one consecutively and one simultaneously. The source texts were two video recordings produced for this study. In both cases, the source speech was delivered under the same recorded conditions (setting, speaker, length of text, adjusted pace, sound conditions). The students made audio recordings of their own interpretations that were subsequently analyzed. The notes used for consecutive interpreting were collected as well. That resulted in 18 audio recordings and 9 sets of notes. The consecutive and the simultaneous interpretations took place in groups and on two different dates, with a span of about two months in between. This was due to practical reasons: it was impossible to gather all participants at one time and a limited amount of time was available for each session.
Great importance was attached to an equivalent difficulty for both source texts, appropriate to the students’ experience with both interpreting modes. During the research study, the participating students clearly had much less experience with simultaneous interpreting than consecutive interpreting. Since the complexity of the source texts needed to match the students’ experience with either interpreting mode, the source text for consecutive interpreting (STC) was more difficult than the source text for simultaneous interpreting (STS). Both source texts were about the same length, but the STC dealt with a more complex and international theme, whereas the STS discussed a human interest story that complemented the STC. Moreover, the STC contained more names and numbers (22) compared to the STS (13). The lexical density of the STC was 12% higher (58.18%) than that of the STS (46.21%). The lexical density was calculated by measuring the number of content words against the total number of words (
Both video recordings opened with a short introduction, which the students did not have to interpret. The STC then required a further 5:26 minutes, and the STS 5:52 minutes, with respective speech rates of 117.05 and 123.40 words per minute. The underlined terms in the source texts (see Appendix 1 & 2) were explained in advance. Furthermore, each student could enter a silent booth to start their consecutive interpretation immediately after having heard the source text.
Some methodological challenges must be mentioned. Firstly, the results (see 4) have a lower ecological validity due to the research setting. The students were asked to interpret as naturally as possible, but all of them were aware that it was part of a research study, although the research objective remained unknown until after the data collection. Also, there was no target audience, as there would be in a real interpretation setting. In addition, participants with more and equal experience in both interpreting modes would increase the validity of the results (cf. Sunnari, 2003). However, using professional interpreters was not feasible within the framework of this study. On the other hand, research using students provides direct insight into strategies, obstacles, and approaches specifically for students. This may inform potential changes within teaching practice. A third methodological challenge lies in the determination of the equivalent difficulty of both source texts, according to the participants’ experience with these two interpreting modes. Lastly, the participants interpreted two video recordings and no live speech, although it may be assumed that this factor had little or no negative impact on the validity of the research (Pöchhacker, 2016).

3.3 Analysis

The audio recordings of the consecutive and simultaneous interpretations were transcribed and consequently compared to the original English source texts. As stated earlier, the omissions were categorized following Kopczynski’s (1980) approach: they were classified depending on the phrases being influenced by the omission (see 2.2.2 and 2.2.3). In this research, ‘phrase’ refers to the linguistic phrase that was being primarily influenced. Kopczynski (1980) sometimes opts for an overarching phrase type that is being affected secondarily. This may result in a changed view of the mental linguistic analysis of the source text by the interpreter (see 2.2.2). This aspect and the descriptive nature of Kopczynski’s approach imply that in this study different omission categories may have occurred. A separate category for the omission of complete and unfinished sentences was adopted. Also, as described under 2.2.3, only non-functional omissions were taken into consideration.
The actual data analysis was carried out as follows. The STC and STS served as a starting point. Both source texts were analyzed to identify all the essential information: that is, all information units that would entail a loss of potentially relevant semantic content when omitted in the interpretation. Those units of essential information were then classified according to their covering phrases, with an additional distinction between core units and modifier units.
Subsequently, the consecutive and simultaneous interpretation transcriptions of every student were compared to the essential information selected in the corresponding source texts. A detailed picture of all non-functional omissions per student per interpreting mode was thus constructed, classified in accordance with the affected phrases (core/modifier). These numbers can be found in the data table in the table rows ‘S1’ to ‘S9’. Some examples of how this analysis was carried out:
Source text for consecutive interpreting: Today, about 700 million people live in extreme poverty.
* noun phrase: core = poverty, modifier = extreme
Interpreting student (Dutch): Today, about 700 million people live in poverty.
* omission of the phrase’s modifier
Source text for simultaneous interpreting: So it’s down-cycled, it’s turned into lesser things.
* verbal phrase: core = is down-cycled, modifier = /
Interpreting student (Dutch): So it’s actually turned into smaller things.
* omission of the phrase’s core
For every student, the omission percentage of all essential information per category was calculated. The relationship between the number of omissions per phrase type (core/modifier) in a participant’s interpretation, and the entire essential information per phrase type (core/modifier) in the source text was measured. An overview of all average omission percentages per interpreting mode can be found in the section ‘Results’ (see 4). They were calculated according to the formula for the arithmetic mean. An example:
The overall mean omission percentage for the noun phrase cores in consecutive interpreting [PS = omission percentage of student]:
(PS1 + PS2 + PS3 + PS4 + PS5 + PS6 + PS7 + PS8 + PS9) / 9 = overall mean omission percentage
However, it is important to put those results into perspective. Some omission percentages had a very high absolute value, but a far less significant relative value. This was often due to the fact that the core or modifier type in question had a very small share of the entire essential information. The lower the total number of essential cores/modifiers of a particular phrase, the greater the impact on the omission percentages when such an element was omitted. In this way, some percentages increased without possessing a real linguistic value.
A further methodological challenge of the analysis was that an ‘error’ or ‘non-functional omission’ on paper was not necessarily perceived as such from the spoken interpretation presented to the listeners (Gile, 1999). Therefore, this report mentions only those non-functional omissions that were important enough to be perceived when listening to the interpretation, from the perspective of someone who has mastered both languages. A certain degree of subjectivity was inevitable.
Only the content of the audio recordings of the interpretations and of the video recordings of the source texts was transcribed, given the research parameters. Aspects such as intonation, volume, stress, accelerations or decelerations, pauses, and irrelevant slips were not taken into consideration. The transcriptions did not serve as a substitute for the original recordings.

4. Results

This overview of all average omission percentages per interpreting mode per phrase (core/modifier) is based on the individual analysis of all transcriptions of the consecutive and simultaneous interpretations. The percentages were calculated according to the formula for the arithmetic mean (see 3.3).
The following phrases were affected by omissions: noun phrase (NP), verb phrase (VP), prepositional phrase (PP), conjunctive phrase (CP), adjective phrase (AP), and adverbial phrase (AvP). In addition, other omissions (OO) and the omission of complete sentences (SO) were also taken into account. The category other omissions contains all omissions that did not fit into one these phrase types. Table 2 shows students’ omissions compared to the total number of omissions possible.
Table 2: Students’ (S) omissions compared to the total possible number of omissions (Tot.).

Consecutive interpreting
Tot. 93 80 38 15 17 17 8 8 4 5 6 4 6 28
S1 17 16 9 2 7 6 1 / / / / / 2 /
S2 25 26 5 2 6 6 2 / 1 1 3 1 2 4
S3 22 22 14 3 6 7 5 / / / 2 2 2 1
S4 12 14 8 1 7 6 1 / / 1 1 / 2 1
S5 18 13 8 / 3 3 2 / / / 3 2 2 4
S6 18 30 9 2 6 6 4 1 2 / 1 1 2 /
S7 13 13 8 1 3 3 2 / / / 1 1 2 2
S8 20 16 10 2 8 7 3 / / 1 1 1 2 2
S9 20 19 10 1 3 3 2 / 1 / 1 / / 5
Simultaneous interpreting
Tot. 101 63 58 35 12 12 11 12 2 6 4 2 4 37
S1 7 11 2 / / / / / / / / / / 1
S2 9 9 2 / 2 3 / / / / 1 / / 4
S3 9 5 7 1 1 / / / / / / / / 1
S4 6 3 2 / / / / / / / 1 / / /
S5 6 11 8 2 1 1 / / / / / / / 2
S6 4 3 2 / / / / / / / / / / /
S7 6 9 4 3 1 1 / / / / / / / 2
S8 4 3 1 1 / / 3 1 / 1 1 / / /
S9 11 8 6 3 1 1 / / / 1 / / / 1

NP = noun phrase; VP = verb phrase; PP = prepositional phrase;
CP = conjunctive phrase; AP = adjective phrase; AvP = adverbial phrase;
OO = other omission; SO = omission of  complete sentences; C = core; M = modifier
Table 3 shows the mean values for the number of omissions per phrase type in CI and SI renditions.
Table 3: Mean values for the number of omissions per phrase type in CI and SI renditions.

Consecutive Simultaneous
Phrase Core Modifier Phrase Core Modifier
NP 19.66% 23.47% NP 6.82% 10.94%
VP 23.68% 11.67% VP 6.51% 5.71%
PP 32.03% 30.72% PP 10.00% 12.50%
CP 30.56% 12.50% CP 27.27% 8.33%
AP 33.33% 20.00% AP / 16.67%
AvP 27.08% 33.33% AvP 25.00% /
OO 33.33% OO /
SO 9.69% SO 4.95%

As stated earlier, the absolute and relative values here must be weighed carefully (see 3.3) using the data table (see Table 1).

5. Discussion

The findings discussed here only apply to the particular group of interpreter trainees who participated in the study (see 3.1). Other possible limitations of the study are discussed in section 3.3. In the case of consecutive interpreting, I compared interpreted renditions with students’ notes, however it may be difficult to link a certain error to a certain cause, because reduction of quality can happen ‘at a distance’ in interpreting. This means a specific trigger may affect a distant speech segment that in itself is not a problem (Gile, 2009).
The large discrepancy between the degree of omissions in consecutive and simultaneous interpreting renditions in this study may be attributable to the fundamentally distinct nature of each interpreting mode. In simultaneous interpreting, the cognitive load is higher because more cognitive tasks have to be performed at the same time. In consecutive interpreting, on the other hand, the interpreter is confronted with the need to rely heavily on memory (see 2.1.1). In the consecutive mode, the interpreter has to keep all successive passages in mind, processing them, until the speech segment has ended. Therefore, (s)he uses note-taking as an aid to compensate for the shortcomings of the memory and processing capacity (Ilg & Lambert, 1996). However, it is impossible to capture all source text information in notes. When the notes of all individual students were compared to their respective omission analyses, almost 90% (88.23%) of all omissions in consecutive interpreting were also omitted from the students’ notes. Thus, the main issue was that the students did not capture the omitted elements, either in their notes or their memories, even after considerable training (see 3.2). The high cognitive demands in simultaneous interpreting did not make this interpreting mode more prone to omissions than consecutive interpreting, even after a shorter training period (see 3.1). The memory factor may have played a larger role than initially thought (cf. Gile, 1997; Gile, 2001b).
Another finding is that for both interpreting modes approximately 4% more noun modifiers were omitted than noun cores. This result suggests that the interpreter trainees tried to preserve the core message when they experienced difficulties processing the information. It is possible that interpreters will focus on the essential information, and more specifically the core issues within the essential information, when their cognitive processing capacity nears saturation. That way, ‘secondary’ elements such as modifiers may be omitted first. That 4% contrast within the noun phrases was similar to Kopczynski’s findings (see 2.2.2) and to the assumptions of other authors (Gile, 2001a; Shlesinger, in Pöchhacker, 2016). In the consecutive interpreting results, the opposite seemed to be the case for the verb phrases. The number of affected cores was much higher than the number of affected modifiers. However, the relative value of the verbal modifiers within consecutive interpreting was too low to draw any conclusions. In the simultaneous interpreting results, both the verbal cores and modifiers were represented in the total of essential information in the source text. Differences in meaning caused by omissions were barely noticeable.
In the consecutive interpreting, approximately one-fifth of all essential noun cores/modifiers and verbal cores were affected by omissions. That is one-fifth of the phrase types that have the biggest share of the essential information (see Table 1). Other phrase types have a rather low relative value due to their very small share in the total of essential information (see 3.3).
This research only focused on interpreting trainees. Professional interpreters are presumably more proficient at distinguishing between core and modifier. Therefore, the same research parameters using professional interpreters might result in a larger contrast. It was clear that all participants achieved the mean values for the most significant phrase types NP and VP, despite initial differences in previous education and English proficiency.
There was a significant difference between the interpreting modes. The total omission average (i.e., the overall arithmetic mean of all omission percentages of all individual students) was 19.51% for consecutive interpreting, and 4.13% for simultaneous interpreting. In other words, consecutive interpreting was 15% (15.38%) more prone to omissions than simultaneous interpreting. This finding aligns with Kopczynski’s research findings: a contrast of 20% between interpreting modes at phrase level (see 2.3.2). These results underline the importance of more research in this interpreting area, if interpreter educators are to help student interpreters achieve more accurate renditions in CI and SI.

6. Conclusion

One of the aims of this research was to find out more about the mental linguistic analysis of the source text, which is carried out unconsciously by the interpreter. It is not possible to make firm statements about the mutual relationship between the numerous linguistic phrase types, because of the difference in their relative values (core/modifier) based on this small study.
The divergence between consecutive/simultaneous modes may have its origins in the cognitive load which is different in consecutive and simultaneous interpreting. Additionally it may concern the strategies interpreters apply to prevent cognitive saturation while retaining the core of the source text message.
This study has shown that student interpreters omitted more information when interpreting consecutively than when they were interpreting simultaneously. Awareness of this tendency may lead interpreter educators to offer trainee interpreters more consecutive interpreting/simultaneous interpreting exercises and perhaps asking students to reflect on what information they thought they had omitted and why. Identifying possible factors which play a role in omissions may help educators develop tailor-made exercises to help students improve.
More research needs to be carried out to explore or confirm the findings of this study. Further research may involve deliberately manipulated source texts to find out more about the mutual relationship between all linguistic phrase types. This may allow researchers to draw conclusions about the mental linguistic analysis that interpreters may be performing unconsciously. Within an educational setting, researchers may experiment with new interpreting strategies and exercises. They could teach and assess note-taking technique to see if students can achieve a more complete interpretation in the consecutive mode. They could also ask students to self-assess whether they have preserved as many modifiers as possible in addition to the cores in both interpreting modes.


Albl-Mikasa, M. (2007). Notationssprache und Notizentext. Ein kognitiv-linguistisches Modell für das Konsekutivdolmetschen. Tübingen: Gunter Narr.
Altman, J. (1994). Error analysis in the teaching of simultaneous interpretation: A Pilot Study. In S. Lambert & B. Moser-Mercer (Eds.), Bridging the gap: Empirical research in simultaneous interpretation (pp. 25-38). Amsterdam/Philadelphia: John Benjamins.
Andres, D. (2002). Konsekutivdolmetschen und Notation. Frankfurt am Main: Lang.
Barghout, A., Rosendo, L.R., & García, M.V. (2015). The influence of speed on omissions in simultaneous interpretation: An experimental study. Babel, 61(3), 305-334.
Barik, H. C. (1971). A description of various types of omissions, additions and errors of translation encountered in simultaneous interpretation. In S. Lambert & B. Moser-Mercer (Eds.), Bridging the gap: Empirical Research in Simultaneous Interpretation (pp. 121-137). Amsterdam, the Netherlands: John Benjamins.
Barik, H.C. (1973). Simultaneous interpretation: Temporal and quantitative data. Language and Speech, 16, 237–270.
Barik, H. C. (1975). Simultaneous interpretation: Qualitative and linguistic data. Language and Speech, 18, 272-297.
Bühler, H. (1986). Linguistic (semantic) and extra-linguistic (pragmatic) criteria for the evaluation of conference interpretation and interpreters. Multilingua, 5(4), 231-236.
De Groot, A. (1997). The Cognitive Study of Translation and Interpretation. Three Approaches. In J. H. Danks, G. M. Shreve, S. B. Fountain & M. K. McBeath (Eds.), Cognitive processes in translation and interpreting (pp. 25-26). Thousand Oaks, CA: Sage Publications.
Garzone, G. (2002). Quality and Norms in Interpretation. In G. Garzone & M. Viezzi (Eds), Interpreting in the 21st Century: Challenges and opportunities (pp. 107-119). University of Bologna/University of Trieste.
Gerver, D. (1975). A psychological approach to simultaneous interpretation. Meta: Translators’ Journal, 20(2), p. 119-128.
Gile, D. (1983). Aspects méthodologiques de l’évaluation de la qualité du travail en interprétation simultanée. Meta, 28(3), 236-243.
Gile, D. (1995a). Regards sur la recherche en interprétation de conférence. Lille: Presses universitaires de Lille.
Gile, D. (1995b). Fidelity assessment in consecutive interpretation: An experiment. Target, 7(1), 151-164.
Gile, D. (1997). Conference Interpreting as a cognitive management problem. In J. H. Danks, G. M. Shreve, S. B. Fountain & M. K. McBeath (Eds.), Cognitive processes in translation and interpreting (pp. 196-214). Thousand Oaks, CA: Sage Publications.
Gile, D. (1999). Testing the Effort Models’ Tightrope Hypothesis in simultaneous interpreting – A contribution. Hermes, 23, 153-172.
Gile, D. (2001a). Consecutive vs. Simultaneous: Which is more accurate? Available at:
Gile, D. (2001b). The role of consecutive in interpreter training: A cognitive view. Available at:
Gile, D. (2008). Local cognitive load in simultaneous interpreting and its implications for empirical research. Forum, 6(2), 59–77.
Gile, D. (2009). Basic concepts and models for interpreter and translator training. Revised edition. Amsterdam/Philadelphia: John Benjamins.
Grbic, N. (2015). Quality. In F. Pöchhacker (Ed.), Routledge Encyclopedia of Interpreting Studies (pp. 333-336). New York: Routledge.
Ilg, G. & Lambert, S. (1996). Teaching consecutive interpreting. Interpreting, 1(1), 69-99.
Jones, R. (1998). Conference interpreting explained. Manchester: St. Jerome Publishing.
Kahane, E. (2000). Thoughts on the quality of interpretation. Available at:
Kalina, S. (2005). Quality assurance for interpreting processes. Available at:
Kopczynski, A. (1980). Conference Interpreting: Some linguistic and communicative problems. Poznan: Uniwersytet im. Adama Mickiewicza w Poznaniu.
Kopczynski, A. (1994). Quality in conference interpreting: Some pragmatic problems. In S. Lambert & B. Moser-Mercer (Eds.), Bridging the gap. Empirical research in simultaneous interpretation (pp. 87-100). Amsterdam/Philadelphia: John Benjamins.
Korpal, P. (2012). Omission in simultaneous interpreting as a deliberate act. In A. Pym & D. Orrego-Carmona (Eds.), Translation Research Projects 4 (pp. 103-111). Tarragona: Intercultural Studies Group.
Kurz, I. (2001). Conference interpreting: Quality in the ears of the user. Meta, 46(2), 394-409.
Kurz, I. (2003). Quality from the user perspective. In A. Collados-Aís (Ed.), La evaluación de la calidad en interpretación: Investigación (pp. 3-22). Granada: Editorial Comares.
Kutz, W. (1997). Gut für wen? Zur Bewertung von Konsekutivdolmetschleistungen. In E. Fleischmann, W. Kutz & P.A. Schmitt (Eds.), Translationsdidaktik. Grundfragen der Übersetzungswissenschaft (pp. 243-254). Tübingen: Gunter Narr.
Matyssek, H. (1989). Handbuch der Notizentechnik für Dolmetscher. Ein Weg zur sprachunabhängigen Notation. Heidelberg: Groos.
Napier, J. (2004). Interpreting omissions: A new perspective. Interpreting, 6(2), 117-142.
Napier, J. (2015). Omissions. In F. Pöchhacker (Ed.), Routledge Encyclopedia of Interpreting Studies (pp. 289-291). London, England: Routledge.
Pio, S. (2003). The relation between ST delivery rate and quality in simultaneous interpretation. The Interpreters’ Newsletter, 12, 69-100.
Pöchhacker F. (1993). Quality assurance in simultaneous interpreting. In C. Dollerup & A. Lindegaard (Eds.), Teaching Translation and Interpreting 2 (pp. 233-242). Amsterdam, the Netherlands: John Benjamins.
Pöchhacker, F. (1994). Simultandolmetschen als komplexes Handeln. Tübingen, Germany: Gunter Narr.
Pöchhacker, F. (2002). Researching interpreting quality: Models and methods. In G. Garzone & M. Viezzi (Eds.), Interpreting in the 21st Century. Proceedings of the 1st Conference on Interpreting Studies (pp. 95-106). Amsterdam, the Netherlands: John Benjamins.
Pöchhacker, F. (2015). User Expectations. In F. Pöchhacker (Ed.), Routledge Encyclopedia of Interpreting Studies (pp. 430-432). London, England: Routledge.
Pöchhacker, F. (2016). Introducing Interpreting Studies. London, England: Routledge.
Pradas Macías, E. M. (2003). Análisis de la fluidez como parámetro de calidad en interpretación. In A. Collados-Aís (Ed.), La evaluación de la calidad en interpretación: Investigación (pp. 97-106). Granada: Editorial Comares.
Seeber, K.G. (2015). Cognitive load. In F. Pöchhacker (Ed.), Routledge Encyclopedia of Interpreting Studies (pp. 60-61). London, England: Routledge.
Shlesinger, M. (1997). Quality in simultaneous interpreting. In Y. Gambier, D. Gile & C. Taylor (Eds.), Conference Interpreting: Current trends in research. Proceedings of the International Conference on Interpreting: What do we know and how? (pp. 123-131). Amsterdam, the Netherlands: John Benjamins.
Sunnari, M. (2003). Expert and novice performance in simultaneous interpreting: Implications for quality assessment. In A. Collados-Aís (Ed.), La evaluación de la calidad en interpretación: Investigación (pp. 235-248). Granada: Editorial Comares.
Tiselius, E. (2015). Accuracy. In F. Pöchhacker (Ed.), Routledge Encyclopedia of Interpreting Studies (pp. 3-4). London, England: Routledge.

Appendix 1: Transcription of the source text for consecutive interpreting

Good morning everyone, and welcome to the Annual Meeting of the World Bank Group and the International Monetary Fund. I would like to first acknowledge our newest country – the Pacific island state of Nauru, which officially joined the World Bank Group just two days ago. So on behalf of the institution, I’m very happy to welcome the Republic of Nauru to the World Bank Group. At our last gathering, we announced that for the first time the percentage of people living in extreme poverty around the world was projected to fall globally. Today, about 700 million people live in extreme poverty – a reduction of more than one billion people compared to fifteen years ago. But the weakening global economy threatens our progress toward ending extreme poverty by 2030. In the global economy, there are not many bright spots around the world – which I would like to illustrate briefly. The first one is the United States, that is one of the developed countries. Second is India, that is of the group of the middle-income countries. But of course, growth remains weak in Europe, and among emerging economies, Brazil is projected to show negative growth once again. So, keeping this in mind, we have just downgraded our global growth economic forecast this year from 2.9 percent to 2.5 percent. In this period of global economic slowdown, we’re also facing major global challenges – forced displacement, climate change, pandemics, to name but a few. We’re now working urgently and in new ways with partners to find solutions to these issues that affect us all. Let me elaborate on two examples. First of all, forced displacement, for instance. We’re using innovative financial tools right now to fund projects that will create thousands of jobs for Syrian refugees and their Jordanian hosts. We’re also working with Lebanon to make sure all children in the country – Lebanese and Syrian refugees – are going to school. We need to do much, much more to provide hope and opportunity for young people affected by conflict, especially in the Middle East and in Africa. The second example, climate change. We must deliver on the promises made at the Paris climate agreement. And yet we are seeing countries around the world about to sign agreements for the dirtiest source of energy, which is coal. So, we’re working with countries right now to piece together deals that would make renewable energy cheaper than coal. Because of these global challenges and also because of the weak global economy, the demand for our services has never been higher outside of a crisis period. So we project that we’ll provide more than 25 billion dollars in loans this year to middle-income countries – which is more than we had projected, and more than any other four-year span in the World Bank’s history in a non-crisis period. The- this increased demand underscores the importance for donor countries to support our replenishment this year for the International Development Association, IDA. This gives low-cost loans or grants to the poorest countries. I want to give you only four realizations of IDA in the last years alone. IDA has supported the recruitment of- sorry, the recruitment and training of teachers; IDA has also supported ante-natal care for pregnant women; improved water services; and the construction or improvement of roads. IDA also helps developing countries enact policies that promote inclusive economic growth, which attract private sector investment, and invest in people’s education and health. A strong IDA replenishment this year will be essential for us to work on our goals to end extreme poverty and boost shared prosperity. Also, to continue our intense focus on the poorest countries in the world, including in Africa and South Asia, and to broaden our work to tackle these global challenges. Thank you very much. I’ll now take your questions.

Appendix 2: Transcription of the source text for simultaneous interpreting

Ladies and gentlemen, I’m a visual artist. I work with plastic bags, which I cut up and sew back together as my primary material for my artwork. And I’ve been doing that for about 20 years or so now. I turn bags into two and three-dimensional pieces, sculptures and installations. After about the first eight years, some of my work started to break down into smaller bits of plastic. After educating myself a little further about plastics, I realized that this was actually a bad thing. It’s a bad thing that plastic breaks down into smaller bits, because it still remains plastic. And what we’re finding is that a lot of it ends up in the marine environment. My initial reaction — and I suppose this would be a lot of people’s first reaction to learning about this — so my initial reaction was, “Oh my God! We’ve got to do something about this! We’ve got to go out there and we have to clean this up.” So I set up a project to go out with a cargo ship. And my intention was to raise awareness about the issue and to pick up the plastic. The plastic I would chip into little bits and mold into bricks that could potentially be used as building material in underdeveloped communities. I began talking to people, people who were already studying the problem of plastics in the marine environment and upon doing so, I realized that cleaning it up would only be a small drop in the ocean. Every year, about 300 million tons of plastic is generated around the world, and what I actually needed to do was take a step back and look at the bigger picture. And the bigger picture was this: we need to find a way to cut the production of single-use and disposable plastics, which are entering the marine environment every day on a global scale. I also realized I was very angry. I wasn’t just concerned about the plastic that’s floating around in the middle of the Pacific Ocean. No, I was concerned about the plastic in the supermarket. All the food and the beverages that I purchase in the supermarket, they’re all packaged in plastic. I’m concerned about the plastic in the refrigerator, I’m concerned about the toxins in the plastic that end up in our bodies. So, then I came together with a group of other people who were also looking into the issue, and 5 years ago, we created the Plastic Pollution Coalition. So, we have set up many initiatives. One of these initiatives is all about recycling — just pretty much everybody’s final argument about being sustainable and being green: the idea of recycling. Now, you put something in a bin and you don’t have to think about it again. What is the reality of that? Well, in the United States, less than seven percent of our plastics are recycled. And if you really look into it, particularly when it comes to plastic bottles, 5.8 percent of it is only down-cycled, or incinerated, or shipped to China. So it’s down-cycled, it’s turned into lesser things. A glass bottle, when it’s recycled, can be a glass bottle again, but a plastic bottle can never be a plastic bottle again. So this is a big issue. And I’m asking people to think about what the Plastic Pollution Coalition calls the “four R’s”: Reduce, Reuse, Recycle and Refuse. Whenever possible, refuse single-use and disposable plastics. Because there are plenty of alternatives. Like, for example, it’s very easy to use a stainless-steel bottle or a glass bottle, fill that up with water, fill that up with filtered water, whatever, instead of having to purchase plastic bottles with water all the time. I guess what I’m trying to say to everybody here — and I know that you people know a lot about the issue already — but what I’m trying to say is that this is a huge problem for our oceans. And it’s also a problem that we, as consumers, have created, and we, as consumers, can solve it. We can solve this by raising awareness of the issue, by teaching people to choose alternatives. So whenever possible, choose alternatives to single-use plastics. In doing so, we can save our oceans, we can save the planet, and we can save ourselves. Thank you.

Appendix 3: The non-automated mental processes represented in formulas (Gile, 2009)

Simultaneous interpreting (SI) Consecutive interpreting (CI)
Listening stage (1) Reproduction stage (2)
Formula SI = L + M + P + C CI(1) = L + N + M + C CI(2) = Rem + Read + P + C

L = listening and analysis effort, M = memory effort, P = production effort, C = coordination effort, N = note-taking effort, Rem = remembering, Read = note reading

Appendix 4: Collected data from the participating interpreter trainees

Student A-, B-, C-languages Previously took part in the Master in Interpreting
Student 1 A = Dutch
C = English, German
Student 2 A = Dutch
C = English, Spanish
Student 3 A = Dutch
C = English, French, Italian
Student 4 A = Dutch
C = English, French, Spanish
Student 5 A = Dutch
B = English
Student 6 A = Dutch
B = English
C = Portuguese
Student 7 A = Dutch
B = English
Student 8 A = Dutch
B = German
C = English
Student 9 A = Dutch
C = English, French, Italian