Available at

1 Corpus-based interpreting studies 
In this first chapter, I focus on different aspects of corpus-based interpreting 
studies (CIS). Section 1.1 focuses on the early history of the field. In Section 1.2, I 
explain why CIS is still a field in its infancy and why it is more “limited” (Bendazzoli, 
2018: 2) than corpus-based translation studies (CBTS). Section 1.3 then gives an 
overview of the different types of existing interpreting corpora, and their 
characteristics. The different recent research orientations in CIS are explained and 
illustrated in Section 1.4. Section 1.5 then concludes this chapter with interim 
conclusions and with the future prospects of CIS. 
1.1 Early history 
Since the 1980s, corpus linguistics has developed at an accelerated speed. 
Indeed, while the compilation and exploitation of English language corpora still 
dominate corpus linguistic research, corpora of other languages have also become 
available and contributed to the diversity of corpus-based language studies. 
Applying corpus linguistic techniques and methods to translation studies was first 
discussed by Mona Baker who predicted that “the availability of large corpora of 
both original and translated text, together with the development of a corpus-driven 
methodology will enable scholars to uncover the nature of translated texts as a 
mediated communicative event” (Baker, 1993: 243). She then suggested a specific 
research agenda that involved the design and analysis of parallel, bi/multilingual 
and above all monolingual comparable corpora. Her first findings were gathered 
among other contributions in the 1998 special issue of the Translation Studies 
journal Meta edited by Laviosa. In addition to the different types of corpora that 
could be compiled such as suggested by Baker, the special issue highlighted the 
potential benefits of using corpora in translator training (Zanettin, 1998) and the 
advantages and disadvantages of parallel corpora (Malmkjaer, 1998), among 
many other things. During its first computer-aided twenty years, corpus-based 

---

Corpus-based interpreting studies 
 
page 10 
translation studies (CBTS) was mainly focused on product rather than process and 
explored linguistic characteristics of translations as texts, such as the proportion of 
lexical to grammatical words and high- to low- frequency words, word repetition, 
and type-token ratios.
In 1998, it was clear that corpus-based studies on written language were still far 
more advanced than studies on spoken language mostly due to the time-
consuming nature of data collection and transcription and to paralinguistic 
dimensions such as rhythm and intonation used by the speaker (Shlesinger, 1998). 
Today, there is still a gap between CBTS and corpus-based interpreting studies 
(CIS), both in terms of corpus size and availability as well as in terms of the 
number of studies and pedagogical applications (Laviosa, 2002; Zanettin et al., 
2003 Kruger, 2004; Aston et al., 2004 in Bendazzoli & Sandrelli, 2009). The 
challenges that I have mentioned partly explain the gap existing between the two 
disciplines. 
Before talking about CBTS in more detail, it is worth defining what a corpus is. The 
Oxford Concise English Dictionary defines a corpus as a “large collection of written 
or spoken texts”. More specifically, the term corpus is associated with at least four 
characteristics: electronic form, size, representativeness and open-endedness 
(Fernandes, 2006). With the advent of the computer, a corpus nearly always 
implies a collection of texts which are represented under an electronic form and 
can be read and analysed automatically or semi-automatically rather than 
manually (Baker, 1995). The size of the corpus also matters but historically, 
corpus-based studies have always relied on huge amounts of data and therefore, 
the term corpus has frequently been associated with big quantities of data 
extracted from large collection of texts. Nevertheless, in CBTS, a corpus can also 
represent what is known to be “small-scale corpora” (Pearson, 1998). As for 
representativeness, the choice of texts in a representative corpus depends not 
only on the size but also of what the corpus intends to represent (Halverson, 1998; 
Kennedy, 1998 in Fernandes, 2006). Finally, a corpus in CBTS is characterized by 
its open-endedness which refers to the flexibility that the corpus should have in 

---

Corpus-based interpreting studies 
 
page 11 
order to allow researchers to answer specific research questions. Researchers 
can then use the texts of this open-ended corpus for different types of 
comparisons and studies (Olohan, 2004: 48). An accurate way to define a corpus 
in CBTS is to refer to that corpus as “an open-ended body of machine-readable full 
texts analysable automatically or semi-automatically and sampled in a principled 
way in order to be maximally representative of the translation phenomenon under 
examination” (cf. Baker, 1995 in Fernandes, 2006: 89).
There are different types of corpora being used in the descriptive and applied 
branches of translation studies. According to Baker’s (1995) terminology, there are 
three main types of corpora: monolingual comparable corpora, parallel corpora 
and multilingual comparable corpora. In the following paragraphs, I will zoom in on 
monolingual comparable and parallel corpora. On the one hand, a monolingual 
comparable corpus can be defined as a corpus that contains components that are 
collected using the same principles, e.g. the same genres in the same domains 
written at the same period and in the same language. On the other hand, a parallel 
corpus can be defined as a corpus containing source texts and their translations. 
They can be bilingual or multilingual. They can also be uni-directional (e.g. from 
English into French or from French into English only), bi-directional (e.g. English 
source texts with their French translations and French source texts with their 
English translations) or even multi-directional (McEnery & Xiao, 2006). 
In the last thirty years, corpus-based translation studies have taken advantage of 
the fast-growing field of corpus linguistics (CL). As for corpus-based interpreting 
studies (CIS), it is a quite new branch of Interpreting Studies that has started to 
emerge in recent years, building on what has already been done in CBTS. Some 
researchers studying interpreting have been examining performance data but have 
been facing challenges specific to the field. In fact, the conditions of reception and 
the production of the translation make the collection and the transcription of oral 
data difficult for the researcher.

---

Corpus-based interpreting studies 
 
page 12 
It is Miriam Shlesinger who, in 1998, was the first to publish a paper on the 
possibilities to extend the corpus-based approach to interpreting and to use 
already available monolingual corpora to test hypotheses. Shlesinger (2008) refers 
to CIS “as an off-shoot of corpus-based translation studies”. Shlesinger’s 2008 
seminal paper can be considered as the cornerstone of CIS. It is worth mentioning 
that Setton (2011) lists all the interpreting studies based on authentic corpora and 
at least five studies were carried out before the publication of Shlesinger’s paper. 
Indeed, Setton mentions Oléran and Napon’s work (1965) as the first corpus-
based research endeavour and the author also mentions Déjean Le Feal’s (1978), 
Chernov’s (1979), Lederer’s (1981) works as interpreting studies based on 
authentic corpora published before Shlesinger’s seminal paper. Yet, it is important 
to add that these studies were manual, as corpus linguistic tools were not in use at 
the time.