Available at

1.4 Recent research orientations in CIS 
In interpreting, comparisons can be made between source and target speeches, 
and between written texts and interpreted speeches in the same language. These 
comparisons can also focus on specific language pairs. The analysis can then be 
either parallel or comparable (Laviosa, 2012). It is also possible to study 
interpreters’ output as a kind of discourse to point out features that make it 
different from other types of discourse (e.g. speakers working under different 
conditions and in different settings, delivery rate, etc.).
This section focuses on the most recent research orientations in CIS, such as 
interpretese and sociological/ideological aspects. 
Features of the product: ‘interpretese’ 
Large-scale CBTS analyses of features such as collocations, part-of-speech 
distributions and lexical richness in monolingual comparable texts have uncovered 
traces of simplification in translated text - a universal trait of ‘translationese’ (Baker, 
1995; Laviosa 1996, 2000). As for ‘interpretese’, in other words, interpreters’ 
speech (Fumagalli, 2000), it has been found that interpreting is closer to original 
speech than to written translation, linguistically-speaking (Shlesinger, 2008; 
Shlesinger and Ordan, 2012; Kajzer-Wietrzny, 2018). Kajzer-Wietrnzy investigated 
the optional complementizer that (2012, 2018) and compared the frequency of that 
used in native English speeches and in simultaneous interpretations into English, 
and used in native English speeches and in non-native English speeches. In both 
studies, the optional complementizer that is less frequent in English native 
speeches, which shows a higher degree of explicitness in interpretations and in 
non-native speeches.
Factors in performance and quality 
Different variables in interpreting performance have already been explored such 
as topic, speed, disfluencies patterns (see Sandrelli et al., 2007; Bakti, 2009; 
Collard & Defrancq, 2019a; Plevoets & Defrancq, 2018) or accents. 

---

Corpus-based interpreting studies 
 
page 19 
In terms of topic, interpreters in EPIC are, for example, much more fluent when 
interpreting impromptu speeches than political discourses, in all language 
combinations and directions (Sandrelli et al., 2007). As for disfluencies, Collard & 
Defrancq (2019a) investigate different types of disfluencies (filled and silent 
pauses, false starts, lengthening). They show that female interpreters produce less 
disfluencies than male interpreters. Bakti’s paper (2009) states that restarts and 
grammatical errors are the most frequent disfluencies in the examined corpus (of 
simultaneous interpreters working from English into Hungarian). Grammatical 
errors might be caused by a cognitive overload and a higher EVS. Finally, 
Plevoets & Defrancq (2018) focus on the disfluency uh(m) and on four 
informational load indicators, namely lexical density, delivery rate, numbers and 
formulaicity. The results indicate that the interpreters’ cognitive load decreases 
when ST are more formulaic (less uh(m)), but increases with ST lexical density 
(more uh(m)). It is important to mention that these are only examples of what has 
been done in the study of interpreting performance.