Intelligibility in English as a lingua franca – The interpreters’ perspective Verständlichkeit in Englisch als Lingua

This article aims at examining the topic of ELF intelligibility from the interpreters’ perspective. Therefore, the focus is put on listener factors affecting intelligibility in settings typical for interpreting i.e. monologic settings. Data from various intelligibility studies are compared with results from a study that tested an ELF user’s intelligibility in a conference-like ELF setting and examined the influence of listener variables such as background knowledge, familiarity with ELF use or proficiency in English. In this study, an Italian speaker gave an impromptu speech in English to participants who subsequently were asked to answer written questions on the topic. The results showed that listeners with more experience in ELF settings reached the highest score in the test, while participants with specialist knowledge were unable to profit from it. The participants’ English language skills played a rather subordinate role. The findings of this study may prove useful for considerations in interpreter training and can contribute to the development of concrete, evidence-based training methods for interpreters in the interpreting subskill of comprehension.


Introduction
The increasing use of English as a lingua franca (ELF) is a development that has radically changed international communication and has also had a major impact on one of the different forms of transcultural communication: interpreting.
Even though the volume of research in the field of ITELF (interpreting, translation and English as a lingua franca) is still relatively low, several surveys and studies have shown that interpreters often struggle with source texts produced by ELF users and that such input can have adverse effects on interpreters' output (for an overview, see Albl-Mikasa 2018).
Therefore, many interpreting scholars agree that interpreters will have to do better at adapting to changing circumstances and demonstrate their added value in ELF settings if their service is to coexist with lingua franca communication. Hence, practising and future interpreters must be specifically trained in the concrete skills necessary to provide high-quality interpretation in ELF settings, a demand already raised by interpreters as early as 1952 (Herbert) and even more so since the significant increase in the use of English as a lingua franca (Albl-Mikasa 2013;Huh 2017;Reithofer 2010). Nevertheless, apart from a study by Albl-Mikasa et al. (2017), interpreter training in this field has not gone beyond anecdotal evidence of coping strategies for interpreting ELF users or mere suggestions of what might work.
It is widely recognised that interpreting is composed of various sub-skills, often divided into reception (listening and comprehension), transfer and production. Many training approaches spilt these part-tasks up even further and suggest to also practise them independently in a component-skills approach (for an overview, see Setton and Dawrant 2016: 60-63).
Without doubt, one of the biggest challenges and hence a key sub-skill required for interpreters in mediated ELF contexts is comprehension. These contexts are typically monologic settings where one speaker gives a talk and interaction is very limitednot only for the interpreters, but also for the audience. These unidirectional communication settings make it largely impossible to use common ELF strategies. Even though studies show that ELF users do apply proactive strategies such as repetition or paraphrasing even in monologic settings (e.g. Björkman 2011), other essential ELF strategies such as negotiating or coconstructing the meaning, requests for clarification or repetitions are difficult to use by listeners in these types of encounters (Björkman 2011: 960) and virtually impossible for interpreters. If interpreters wish to enhance their comprehension sub-skill for assignments in ELF contexts, evidence-based knowledge about listener-related factors that allow better comprehension of ELF users is essential. Therefore, this paper will explore the concept of ELF intelligibility from the interpreters' point of view and review studies on listener factors affecting intelligibility.
Furthermore, it will present data from a study with a typical conference-type ELF setup characteristic for interpreting contexts with an expert ELF user speaking to a multilingual audience. The study was aimed at assessing the listener variables that affected comprehension by correlating the participants' results in a comprehension test with background variables collected in the studysuch as the listeners' background knowledge, familiarity with use of English or proficiency in English. The results should help identify the factors that facilitate the interpreting part task of reception in an ELF context to then use this knowledge in specific ELF training for interpreters.
Since listening and comprehension are skills that are not unique to interpreting, this paper considers findings from all listening contexts comparable to mediated settings to be valuable for the training and continuous professional development of interpreters regarding the part-task of listening and comprehension.

Intelligibility in ELF contexts
Multiple definitions of the notion of intelligibility have been developed in the field of linguistics. A frequently used definition is the threefold distinction proposed by Smith (1992: 76), who differentiates between intelligibility (recognition of words/utterances), comprehensibility (understanding the meaning of a word/utterance [locutionary force]) and, interpretability (understanding the meaning behind a word/utterance [illocutionary force]).
Smith also assumes that the three levels interact. One may ask, however, whether intelligibility, comprehensibility and interpretability are required to ensure good interpreting performance. In interpreting studies, most scholars base their concept of comprehension on the well-established notion grounded in cognitive psychology that comprehension occurs on the micro and the macro levels: recognition and decoding of linguistic information is combined with prior knowledge and setting-specific facts in order to transfer the speakers' input in a given situation . This would imply that the source text would have to be intelligible and comprehensible. In some instanceswhen interpreters merely transcode the message e.g. in lists or enumerationsintelligibility might suffice. Interpreters often claim to interpret what "the speaker actually wants to say" and thus refer to interpretability, but this is difficult to measure and has thus been largely disregarded in linguistics (Pickering 2006: 220).
In the following, I will explore in greater detail which factors have an influence on intelligibility, which I will use as the umbrella term for the three concepts and indicate when they are used in Smith's sense.

Variables affecting the intelligibility of ELF users
This section will discuss relevant findings on ELF users' intelligibility from different fields of linguistics, some also from outside the ELF paradigm who mainly looked at non-standard English.
While it has been acknowledged that understanding is co-constructed by speakers and listeners (Pickering 2006), research has focussed largely on speaker factors affecting intelligibility.
Among the most frequently studied topics are accent and phonology, but variables such as speakers' lexis, syntax or grammar have also been examined. A broad range of approaches have been used to determine intelligibility: in the field of ELF research, many scholars analyse interactions to identify factors responsible for a breakdown of communication (e.g. Deterding 2013;Jenkins 2000). In some studies, the participants are asked to jointly complete a task using English as a means of communication (e.g. Kennedy 2017; Thir forthcoming). Other analysts use scales to assess listeners' subjective understanding of L2 speakers (Anderson-Hsieh et al. 1992;Munro et al. 2006). Furthermore, word or sentence recognition tasks (Wang and Van Heuven 2014), transcriptions (Kennedy and Trofimovich 2008) or cloze tests (Smith and Rafiqzad 1979) have been employed. Another approach relies on comprehension tests that listeners take after hearing an utterance by an ELF user (Bulatović et al. 2019;Wilang and Teo 2012a); this is the method used in the study described in Chapter 3.
When looking at intelligibility variables of ELF users cited in the field of interpreting studies, accents have often been mentioned as a difficulty for interpreters (Cheung 2015). Some interpreters believe that non-standard phonology (Chang and Wu 2014) or prosody (Lin et al. 2013) and thus intelligibility in Smith's sense are the main problems in ELF contexts. Albl-Mikasa et al. (2017), however, found that the main problem trigger for interpreters was not ELF users' accent by itself, but rather non-standard phonology combined with unorthodox wording and phrasing. Due to the "speakers' constrained pragmatic fluency and restricted ability to express their key point in precise terms […] interpreters find it hard to (quickly) grasp the argumentative logic behind the phrases and expressions" (Albl-Mikasa et al. 2017: 232). This implies that, in Smith's terms, not only intelligibility, but also comprehensibility (and interpretability) pose problems for interpreters.
While it is likely that the level of intelligibility of an utterance is determined more by the speaker than by the listener (Hansen Edwards et al. 2018: 554), knowledge about speaker factors is less relevant to interpreters, since they cannot control or influence a speaker's output. Interpreting researchers' main goal therefore is to find ways to help practising and future interpreters to better understand the ELF users they have to interpret.
Turning thus to listener variables affecting intelligibility in ELF contexts, research has shown that they include a plethora of factors. I will focus mainly on the ones examined in the study that is to be reported upon in this paper.

Background knowledge
The essential role of prior topic-specific knowledge in the comprehension process is widely recognised (McNamara et al. 2007). In ELF studies, the question of background knowledge is closely related to the notion that shared knowledge enables experts from a domain to communicate effectively with each other using ELF in their community of practice. It has been assumed that experts in the field of the study described in Section 3business communicationcan overcome linguistic barriers thanks to their specialisation in the respective field (e.g. Kankaanranta and Louhiala Salminen 2018). Poncini (2004) found a culture of its own at business meetings, a groupness that evens out difficulties arising from the speakers' different levels of English proficiency.
Turning to experimental findings from other relevant fields of linguistics, early studies (Gass and Varonis 1984;Smith 1992) also showed that prior knowledge increased the intelligibility and interpretability of ELF users.
Interpreters' prior knowledge cannot be assumed to equal their clients' expertise, since they are usually not part of the expert groups they work for. Therefore, interpreters try to acquire as much subject knowledge as possible during preparation for a specific task or for domains they regularly work in (Kalina 2005). In an experimental study (Díaz Galaz et al. 2015), thorough preparation and thus acquisition of background knowledge was shown to lead to better quality in interpreting difficult technical segments. Scholars believe that preparation also helps when source text difficulty arises from non-standard input (Albl-Mikasa 2013; Huh 2017; Reithofer 2018) and practising interpreters likewise see it as a potential coping strategy (Chang and Wu 2014).

Knowledge of ELF users' first language
One recurring feature observed in ELF contexts is that speakers creatively apply elements of their first language (L1) also when speaking English (e.g. Guido 2012). Many studies have thus investigated whether intelligibility of ELF users is higher when speakers and listeners share an L1, i.e. whether a German listener understands a German speaker of English better than a Spanish or an American listener would.
In this respect, Bent and Bradlow (2003) proposed the concept of the matched interlanguage speech intelligibility benefit, an advantage resulting from the interlocutors' common knowledge about consonant and vowel categories, emphasis patterns, intonation, etc. Another explanation for this hypothetical advantage could be that persons with the same L1 typically share a common culture in the sense of common knowledge about (linguistic) conventions and the like (Kachru 2008: 311) and thus understand each other better in an ELF context. This advantage was confirmed in various studies in the field of ELF and beyond with very different designs (Jenkins 2000: 57;Lee 2004;Smith 1992;Wang and Van Heuven 2015;Yule et al. 1990). However, Hansen Edwards et al. (2018) found that it depends on the language proficiency of listeners whether a shared linguistic background is beneficial to intelligibility or not. Tauroza and Luk (1996) also failed to find what they call an own accent advantage.
In the field of interpreting, Chang and Wu (2014) as well as Katikos (2015) concluded that interpreters understand ELF users from a common L1 background best and find those with significantly different language backgrounds most difficult to interpret. Furthermore, there is evidence for a benefit of knowing a speaker's L1, that is, not having the same L1, but understanding the speaker's L1 very well. Several interpreting studies (Kraus 2011;Kurz and Basel 2009) examined this phenomenon and all found that knowing an ELF user's L1 well or having it as a working language immensely facilitates the interpreter's task. In reference to Bent and Bradlow, Albl-Mikasa (2013: 105) called this the shared languages benefit.

Familiarity with use of English
Studies from various linguistic sub-disciplines on familiarity with non-standard English will be reviewed in this section. Although not all of them are based on the ELF paradigm, I believe their findings are nevertheless relevant for this paper.
Some linguists believe that non-L1 listeners generally understand other non-L1 speakers better than L1 listeners of the language that is being spoken do. They see the reason for this in the fact that non-L1 listeners are familiar with non-standard use of the language and know and recognise strategies employed by non-L1 speakers. Bent and Bradlow (2003) coined the term mismatched interlanguage speech benefit for this hypothetical phenomenon.
Several scholars looked at how familiarity with non-standard speech in general affects comprehension. It has been observed that ELF users are more intelligible for listeners with previous exposure to non-standard speech (Kennedy and Trofimovich 2008), be it through international experience (Hansen Edwards et al. 2018) or contact with ELF use at work, through (social) media or while travelling or living abroad (Wilang and Teo 2012b).
Regarding familiarity with a specific accent or rather similect (Mauranen 2018), several studies showed that it increases the ability to understand the similect (Derwing and Munro 1997;Munro and Derwing 1999;Wingstedt and Schulman 1987). Smith and Rafiqzad (1979) suggested that this familiarity materialises already after a short time of exposure. Weil (2001) indicates a period of four days, after which even different speakers with a similar accent are more intelligible. Clarke (2004) found that this normalization can become apparent after only a few hours or even after only 1 min or two to four sentences of exposure to the same speaker.
Practising interpreters confirm this and claim they are able to adapt to a specific accent in a few minutes (Chang and Wu 2014: 176-177). Student interpreters reported similar effects (Huh 2017: 73).
Familiarizing interpreters with different similects is frequently mentioned as a potential training strategy for ELF settings. Scholars assume that frequency of exposure allows for the storage of the specific linguistic patterns in long-term memory, which then facilitates the interpreters' listening effort (Huh 2017: 60).
Other linguistic studies (Clarke 2000;Munro et al. 2006), however, rebut the advantage of this type of familiarity or find only weak evidence of a link to intelligibility.

English proficiency of listeners
Another variable that some researchers believe affects the intelligibility of ELF users is the listeners' proficiency in English. Listeners with higher proficiency in English are thought to have fewer problems understanding ELF users. Smith (1992) showed that listeners' language proficiency influences intelligibility of ELF users, but that this does not mean L1 listeners are best able to understand ELF users. Wilang and Teo (2012b) found ELF listeners' exposure to English in education to correlate with comprehension of ELF users. In their experiment, listeners with graduate studies in English were better at understanding ELF users. Matsuura et al. (2017) also found listeners who were more proficient in English to be better at understanding ELF users, since they relied less on the phonological input and were able to use contextual clues.
Turning to ELF settings typical for interpreting assignments, variables such as background knowledge, familiarity with ELF contexts or proficiency in English can be analysed by using data from a study which will be described in the following chapter.

Study
The findings described below are part of a larger research project (Reithofer 2013(Reithofer , 2014 that compared the effect of ELF and interpreting. In what follows, I will only report the part regarding ELF. Unlike most ELF studies with dialogic settings, this study sought to create a context typical of the ones in which interpreters frequently encounter ELF users: a monologic talk by an expert speaker addressing a large (expert) audience with no or little room for interaction. Therefore, a conference-like setting was created where participants with various L1 were asked to listen to a conference-type speech by a proficient ELF user. Afterwards the listeners completed a comprehension test on the content of the talk.

Research question
As mentioned above, interpreters have no influence on the communicative resources of the ELF users they interpret. They can only maximise their level of understanding of the speakers' input. In order to do this, interpreters and interpreting trainers need to be aware of the listener factors that facilitate comprehension of input by ELF users. In the following, the variables that affected comprehension in a conference-type setting will be assessed by correlating the participants' results in the comprehension test with background variables collected in the study to find listener factors that increase comprehension.

Speaker
It has been acknowledged that intelligibility of ELF users can vary immensely (i. a. Wang and Van Heuven 2014). Since no single user can be regarded as the typical ELF user, it was decided to choose a realistic representative of presenters at conferences: a non-native speaker of English who frequently communicates in highlevel ELF settings. After video recordings with several academics at a prestigious Italian university, the author's research team and international interpreting experts chose an Italian professor with extensive experience of lecturing in Englishin his university courses and at international academic conferences. He had been recorded giving an impromptu speech of 15′24′′ without any notes or script on one of his main research topics in the field of marketing. The talk included typical features of authentic lecture discourse and spontaneous speech such as false starts, repetitions, redundancies and rhetorical questions.
The topic turned out to be highly innovative as confirmed in a pre-test with subject-matter experts.
The speaker's accent was rated by 46 expertsinterpreters and ELF scholarson a 7-point interval scale from one (no perceptible foreign accent) to 7 (very strong foreign accent). Eight raters were L1 speakers of English, 38 were non-native speakers of English with various European L1. His accent was given a mean rating of 5.48 (SD = 1.1) and can thus be described as slightly stronger than average. The raters, however, also mentioned that he was very intelligible and had a high level of proficiency in English. Nearly all raters (93.5 %) identified him as an Italian speaker of English.

Questionnaire
The questionnaire administered to the participants consisted of two parts: one with listening comprehension items and the other with background questions on the participants and subjective assessments of various parameters. The listening comprehension test comprised multiple choice (MC) and (half-)open-ended questions that were based on the wording of the speech. Lexical overlap was explicitly desired to increase the possibility of recognition of information and to lower the need for recall, which would have tested the participants' memory capacity rather than comprehension (Berne 1993: 24). It is true that when information is queried directly after listening, not only the intelligibility of lectures is tested, but also the memory and retention of the test subjects. Since the process of understanding cannot be observed, however, the test subjects had to be given some kind of task. The participants were comparable regarding their age, education and study situation. They were parallelized according to their overall grades in order to distribute students as equally as possible, which was intended to ensure that the influence of the factor memory was kept as small as possible.
Eight out of 11 items were MC questions. The sequence of correct answers and distractors was changed from item to item in order to avoid format effects.
The three semi-open questions were phrased in such a way as to make the correct answers clear from the beginning (e.g. "Name the speaker's three concrete examples for customers with a fragmented identity"). They were thus not fully open-ended questions, but, compared to closed questions, had the advantage that the right answer could not be guessed. A calculation of the level of difficulty of all items showed that they were not the most difficult and thus did not distort the results. All items were of medium difficulty, i.e. between 0.2 and 0.8, where 0 is the most difficult and one the least difficult.
The questionnaire was tested beforehand in cognitive interviews with a small number of business experts, as recommended by Willis (2005), to evaluate its suitability and intelligibility.

Participants
In this paper, we will focus on 67 participants out of 139 in the original study, since those 67 listened to the ELF user (while the others listened to the interpreter). We will disregard the 72 subjects who listened to the interpretation not relevant for this paper.
Among those listening to the original speaker were 42 business students at two Austrian universities of applied sciences (Fachhochschulen, referred to as FH1 and FH2) who have to pass highly competitive admission tests. These specialist students can be considered a realistic target audience for the presentation by the university lecturer. 25 participants were interpreting students at the Vienna University's Centre for Translation Studies (CTS) and thus non-experts in the field of marketing. All of the participants had L1s other than English (cf. Table 1).
The participants were comparable in terms of age (cf. Table 2).
As mentioned before, listening and comprehension are part-tasks of, but not skills unique to interpreting. Thus, findings from all types of listeners in ELF contexts that are comparable to mediated settings are relevant for gaining deeper insights into the variables affecting intelligibility in ELF contexts that could be used in the training for the interpreting sub-skill of reception.

Procedure
The conference simulation for the FH1 and CTS groups took place in a lecture hall at the Centre for Translation Studies. The questionnaire was handed out to the participants as soon as they entered the room, but they were told not to read it until instructed to do so. The procedure was then explained in detail. During the talk, students saw the original speaker on a large video screen. After the speech they were asked to complete the questionnaire.
The test with the FH2 group took place on the premises of that institution, with comparable equipment and facilities, and followed the same procedure.

Results
The effect of the variables of subject-matter knowledge, familiarity with ELF use or the speaker's similect and language proficiency on intelligibility was investigated on the basis of inter-group comparisons as well as correlations between the score in the comprehension test and biography-related listener variables.

Background knowledge
Background knowledge or prior knowledge of the topic also referred to as subjectmatter expertise has frequently been cited as a facilitating factor for ELF intelligibility (see Section 2.1.1). The business students were expected to share relevant knowledge with the speaker and thus understand him better than the interpreting students. Apart from their expertise resulting from their field of study, FH1 and FH2 students also considered themselves to be more familiar with the subject of the speech than CTS students. In their responses to the question "How familiar were you with the topics discussed in the speech? (from 1 = very familiar to 7 = not familiar at all)", there was a two points difference between the median of CTS and FH1 students and a one point difference between the CTS and the FH2 groups (cf .  Table 3).
Nevertheless, the results of the comprehension test for the three groups showed that the non-experts from CTS had reached the highest score at 8.3, compared to 8.1 by FH and 5.7 by FH2 (cf . Table 4). Thus, even though FH1 and FH2 students participated in a course in the domain of the speech and claimed to be more familiar with the subject, the interpreting students achieved better results in the comprehension test.

Familiarity with ELF use
Several studies indicate that listeners who have more experience in ELF settings and with ELF users of any linguacultural background also find them more intelligible (see Section 2.1.3). Therefore, in our study, the participants were asked "How often do you speak English with persons who use English as a foreign language?". 1 The answers (very often, often, sometimes, rarely, never) were grouped into two categories: participants with more contact with ELF users (very often, often), and those with less (sometimes, rarely).
In all test groups, respondents in the "more experience" category achieved better test results: in FH1 by 2.6 points, in FH2 by 2.1 points and in CTS by one point (cf . Table 5). A t-test showed a significant difference in the FH1 scores for less contact (M = 6.82, SD = 2.4) and more contact (M = 9.43, SD = 3.6); t (25) = 2.24, p = 0.035. Because of the small number of participants in FH2, no t-test was conducted, but the 2-point difference is a highly meaningful descriptive outcome. Hence, the participants who spoke with ELF users (very) often also understood the Italian speaker better.

Familiarity with the speaker's ELF similect
Another variable that, according to several studies, influences the intelligibility of ELF users is the listeners' familiarity of the speakers' specific accent or similect, in 1 I am aware of the fact that EFL and ELF are very different concepts. The question was enquiring about interactions with "ELF users with another L1." Since the questionnaire was addressed to mainly non-linguists, I assumed that the participants would not understand what is meant by "ELF users with another L1." ELF intelligibility for interpreters our case Italian English. For this reason, respondents were asked to indicate (on a 5-point ordinal scale) how often they had heard Italians speak English.
In the FH1 group, 19 participants reported prior experience with Italian users of English. Surprisingly, the participants who stated that they had never heard an Italian speak English achieved the highest test score. This is in stark contrast to the findings discussed in Section 2.1.2. On the other hand, for those with experience, the test score increased with more experience. The results are, therefore, contradictory. It should be noted, however, that the group sizes were very small, and no one had chosen the answer option "very often".
The FH2 and CTS groups did not show a clear picture either. No linear correlation was found in scatter plots. Due to the very small group sizes, no inferential statistics were carried out.
Thus, in this study, familiarity with the specific accent could not be confirmed as a factor positively influencing intelligibility.

Listeners' L1
Overall, the number of participants grouped by their L1 is too small for general statements or statistical calculations about the influence of L1. They will be presented here to show a certain trend, but cannot necessarily be extrapolated.
Almost all students in FH1 and FH2 indicated German as their L1. In the CTS group, five test participants shared the first language of the speaker -Italian. However, they achieved a lower test score than the other listeners. The shared languages benefit therefore did not materialise (cf. Table 6).
Participants who did not indicate German or Italian as their first language achieved the highest score. It is not known if they were foreign students living in Austria or Austrians with a different L1. In both cases, it can be assumed that these participants were more used to creatively using their multilingual repertoires and might thus have been better at understanding such strategies employed by the speaker.

English proficiency
The influence of listeners' English proficiency on the intelligibility of ELF users was tested using various results. First, it was evaluated based on the participants' selfassessment of their English skills. In the questionnaire participants were asked to rate their English listening skills on a 7-point scale from 1 = very good to 7 = very poor. The results indicate good, above-average listening skills (cf. Table 7). The self-reported English skills of CTS and FH2 participants showed only a low to very low correlation with their test scores. For FH1, however, results of a Spearman correlation indicated that there was a significant moderate correlation between self-assessment of language skills and the test score (r s = −0.47, n = 27, p = 0.007). Thus, in one of the groups, participants with a better knowledge of English were able to achieve better results.
Furthermore, the influence of English proficiency on the score was assessed using additional data. Some of the respondents from FH1 attended a bilingual version of their MA programme with a strict selection procedure at the beginning of the course and many classes in English. One can assume that they were more proficient in English than the other FH1 students. A comparison of their mean values in the listening comprehension test shows that the score of the bilingual students was more than half a point above that of the German-speaking students (cf . Table 8). However, this difference was not statistically significant.  These results indicate that in this study, the English proficiency of the participants had only a limited effect on the comprehension test scores and their understanding of the ELF user.

Conclusion
The goal of this paper was to gain knowledge about ELF intelligibility that would be relevant for interpreters, in particular for the interpreting sub-skill of comprehension in ELF settings. Since interpreters often identify ELF users as a challenge, the article first reviewed findings regarding the intelligibility of ELF users. The main focus was on listener variables affecting ELF intelligibility, i.e. variables that interpreters, in their capacity as intermediary listeners, themselves can influence in order to improve their performance in ELF settings. Furthermore, the paper presented results from a study that explored intelligibility factors in a conference interpreting setting, i.e. a specialised event with monologic communication that did not offer the possibility to apply common ELF communication strategies such as the negotiation of meaning or clarification procedures.
Since listening and comprehension are skills that are not unique to interpreting, the paper considered findings from all listening contexts that resemble interpreting settings to be relevant for the analysis.
In the author's study with 67 participants, frequent interactions using ELF and thus experience in ELF settings were found to be a listener variable that strongly affected intelligibility. This finding is in line with studies in various fields of linguistics, where several research reports have shown that listeners' familiarity with ELF use increases comprehension (see Section 2.1.3). Nevertheless, in this study, familiarity with the speaker's specific accent did not positively influence intelligibility.
Surprisingly, the variable of background knowledge or expertise was not found to have a strong impact on intelligibility. In the comprehension test, nonexpert (interpreting) students achieved higher scores than the students from the speaker's field of expertise, even though the former stated to be less familiar with the subject. This suggests that the mere fact of belonging to a certain group of experts or having a general knowledge of a topic does not automatically guarantee better understanding of innovative sub-topics such as the one presented by the Italian speaker in the study and that only very specific prior knowledge leads to intelligibility benefits. Thus, in monolingual ELF communication situations, groupness seems to require more than the mere affiliation to a discipline.
Furthermore, research has shown that professional interpreters possess highly efficient comprehension strategies that allow them to carry out a deeper semantic analysis than untrained bilinguals or native speakers. This might imply a more automatic way of performing linguistic analyses in interpreters (Yudes et al. 2013). It is possible that the non-expert interpreting students had already developed better listening comprehension skills than the business students and were thus able to outperform the expert listeners.
This might also imply that interpreters with a high level of specific preparation for a technical conference could be able to better understand ELF users than the general (expert) audience and thus provide an added value with their service.
Another explanation for the non-experts' good performance might be that many of them did not have German as their L1, which means that they were either foreign students living abroad or Austrians with a different L1 than German. This might have increased their awareness of strategies employed by speakers with multilingual repertoires and might have helped them offset the expert students' subject knowledge edge.
Other listener variables, such as participants' English proficiency, only had a moderate influence on comprehension.
As for the variable of a first language shared by speakers and listeners that is often mentioned as a factor facilitating comprehension, only five participants had the same L1 as the speaker and thus no valid conclusions can be drawn in that respect.
It is evident that further research into ELF intelligibility in interpreting settings is needed. The findings of the study do, however, suggest that training of the interpreting part-task of reception in ELF contexts should focus on a maximum of exposure of (future) interpreters to ELF settings and users. In training programmes, a specific emphasis might be put on comprehension tasks and interpreting exercises with non-standard accents and similects that are likely to be found in students' future career settings with regard to regional specificities, markets and employers such as international organisations.
The findings suggest that in mediated ELF contexts, very specific and in-depth knowledge of the topic in question might be necessary. Thus, the students' research and preparation skills should be geared towards an extremely targeted way of dealing with preparatory documents.
There is every reason to believe that interpreters will encounter an everincreasing number of ELF users as speakers in their work. Thus, interpreting studies must seek to develop an evidence base for didactic approaches to interpreting in ELF contexts. It is to be hoped that the results of this study will bring us one step closer to this objective.