ELF and intelligibility in European HE
University contexts in Europe comprise monolingual institutions and historically bilingual institutions working and teaching in two languages. They vary in the way additional languages affect the institutional linguistic ecology (see e.g. Veronesi and Nickenig 2009). Individual additional language socialisation for academic purposes used to be an asset, but from the 1970s a 'BTA' (Been To America, cf. Swales 2004: 2) became an informal language requirement for a career in the sciences. Research on how university language practices in Europe have evolved in the last 20 years, including ELF, is found in Haberland et al. (2008), Smit and Dafouz (2012), Haberland et al. (2013) and Doiz et al. (2013).
In Switzerland, academic language skills beyond L1, particularly English, are increasingly required even at Bachelor level. Politically, this is contentious and may disadvantage individuals who studied other national languages at school before English. On the other hand, its use is accelerated by the non-territorial nature of ELF, the dynamic complexity and availability of ELF as a resource (cf. also Hulmbauer 2011) and by earlier, or more varied mobility, faster exchange of information, more ad-hoc languaging for communal disciplinary speech events and more multilingual fluctuation in the language repertoires of local academic communities. It is often overlooked that it is stable, spatially-bound communities, not malleable constellations of interlingual speakers (Hulmbauer 2013), that develop code switching and mixing, or parallel or mediated-simultaneous or unreciprocated language use (cf. also Hulmbauer 2011: 44). Interpersonal encounters involving mobilisation of many different resources are often cited as exemplifying multilingualism because many codes are visible on the surface. Hulmbauer's (2013) and Cogo's (2012) deeper approach allows constellations of intercultural speakers and the intralingual and interlingual resources that flow into ELF encounters to reveal their multilingual saturation.
Returning to specific academic ELF settings, it should be noted that, unlike in the locally motivated, public disciplinary lunch-time speech events this author examined (Schaller-Schwaner 2012: 148-158), consistent ELF use is not predictable from the official labelling of a programme as English-medium. Varying patterns of moment-to-moment local language use in courses nominally taught in English were found in certain Swedish, Danish and Norwegian settings of Bologna-designed 'English-medium' international university courses
(cf. Neville and Wagner 2008, Ljosland 2011, Soderlundh 2013). This is balanced by ample evidence from other Nordic university settings operating in ELF (e.g. Bjorkman 2013, Hynninen 2013, cf. also Mauranen 2003, 2012), indicating that findings from UFR may be useful for others.
At UFR, ELF is as complex and dynamic as elsewhere, but is at times the weakest language locally with the lowest number of fluent speakers and the least institutional support. It may be the weakest link in a student's multilingual repertoire. As an embedded language (Schaller-Schwaner 2008b: 264-265; but cf. Hulmbauer 2011: 44-48), it can be difficult to maintain as an output language. Speakers of oral presentations know that they share other or stronger languages with some members of their audience but also that, if they do not stay in 'codesharing' lingua-franca mode (Schaller-Schwaner 2010, 2011; cf. Hulmbauer 2011), they will exclude or alienate others, as explained in Schaller-Schwaner (2011: 436), splitting the bilingual audience into addressees and auditors (Bell 1984, 2001) and flouting the genre expectations.
This means that ELF as a practice needs getting used to. Linguistic insecurities or classroom roles such as 'doing being a student' (e.g. Attenborough 2011) need to be tackled in the process of becoming a confident and aware ELF user. Developing multilingual awareness for ELF users means presenters not only having to cope with competing responses from their own complex language repertoires, learning to override them and sticking to English, but also having to deal with language contact phenomena within ELF and monitoring their own output.
Many ELFA users originally learn English as a foreign language at school in largely same-language groups. They are thus used to typical language contact features connected to their own particular linguistic backgrounds and usually find the English of speakers with similar multilingual repertoires easier to process.
Roughly, this corresponds to Mauranen's (2012: 29) notion of "similects", L2 lects of English that "arise in parallel, not in mutual interaction" but do not develop further as speakers of the same L1 have no reason to interact in English. However, at UFR, speakers do interact in ELF to include someone who does not share their L1, to signal disciplinary membership, or to socialise doctoral students into an English-medium spoken academic genre.
For the study of ELF in European HE, the empirical work on ELF intelligibility by Jenkins (2000) remains pivotal. Researching mutual intelligibility and accommodation in interactions among non-L1 users of English in terms of features and processes, Jenkins (2000) proposed the Lingua Franca Core, based on what caused misunderstandings in English among internationals and featuring certain stress patterns, consonants and consonant clusters, vowel lengths, and one central vowel. Jenkins also highlighted that interpersonal accommodation processes and adaption of one's pronunciation to high-stakes situations were crucial for ELF. Jenkins' analysis led to disputes but also to explorations and applications (e.g. Walker 2010) as well as to further research in this area, such as Deterding (2012), which also focuses on (non-overt) misunderstandings when there is no breakdown of communication or no interference with overall comprehension. Partial intelligibility surfaces in the analytical process when a transcriber does not manage to recognise a lexical item or does not hear a word accurately (Deterding 2012: 187).
As regards the ELF academic presentations studied here, partial or nonunderstanding is usually not overt either. As there is no visible interaction, absence of understanding can go undetected: on the one hand, identifying words in running speech that one cannot interrupt is more difficult, while on the other nobody will notice if the listener has a problem. In addition, intelligibility is affected by shared schematic knowledge. Members of a disciplinary community construct meanings more easily during an oral presentation than someone who depends on actually recognising a word because they are simultaneously building disciplinary knowledge. Pictorial information supports cueing into the existing disciplinary schemata and helps construct them but itself involves degrees of familiarity with specialist semiotic conventions (Danielsson 2014).
Intelligibility as a linguistic construct encompasses recognisability of words and a listener's ability to recognise words or stretches of speech and identify them with sense units (Field 2005). Ethnographically, it is also an emic category and an account of experiential data. For an EAP teacher/researcher and ELF listener who is not a member of the discipline, for example, emic intelligibility entails an element of trying to construct disciplinary knowledge, but also an element of difficulty of transcribing, as shown in Deterding (2012), in which unintelligibility surfaced not during ELF interactions but afterwards when an interlocutor as transcriber did not recognise or misidentified words during transcription. In transcribing ("representing spoken ELF in written form", Breiteneder et al. 2006: 161), one realises the extent to which one cannot identify a word reliably or when one cannot decode the sound sequence heard, that is, associate it with a meaning.
While detailed quantitative tests of intelligibility such as Field (2005) can contribute essential insights into intelligibility as an etic category, the emic point here is informants' own (self-)perceptions of strain, listening effort or frustration with themselves or the presenter, which participants associate with pronunciation. Put differently, lack of or limited intelligibility is not a property of the speech signal alone but is intertwined with factors such as speaker, listener, context, purpose and private or shared assessment of a participant experience. As an emic category, it is usually reported and discursively constructed as 'inability to understand' or 'understanding difficulty', and comprises individual and collective experience of language use as well as the sense people make of it, including attitudes, social perceptions and intentions, personal positionings, or even prejudice.
What familiarity contributes to enhancing intelligibility also seems connected to the interplay of linguistic criteria and an individual's expectations, perceptions, and memory. Speech perception and recognition of spoken language appear to be a highly integrated process in which experience with linguistic properties and experience with 'indexical' properties that are specific to speaker- identity are attended to and retained together (e.g. Szakay et al. 2012, Borrie et al. 2013).
To be conceptually clear about understanding (cf. Pitzl 2005, 2015), ability to transcribe can be seen as one possible operationalisation of intelligibility but does not equal understanding, as Deterding (2013: 11) seems to be suggesting. Real-time listeners orient towards content. Gaps or disturbances are not contemplated, as perceptual fluency is essential, but they register as strain or partial understanding. When transcribing or listening to a recording repeatedly, listeners process differently and declaratively; they have time and become aware of what they hear, of what/why they fail to understand, and of perceptual and productive fluctuations. This is pedagogically relevant, as are doubts about intelligibility, which undermine presenters' confidence and impact their delivery. EAP learners' concern about pronunciation and intelligibility in the interactional vacuum of an oral presentation is a pro-active way of managing understanding. Besides, in planned speech, measures to tackle the transience of spoken words (e.g. explicit structure, visuals, gesture, mobilisation of prior knowledge) can be taken.