Introduction
Only a few years ago, legal restrictions made EMI (English as Medium of Instruction) seem improbable. Today, however, EMI is rapidly expanding throughout Flemish universities. This growth raises important questions: Are institutions inadvertently disadvantaging non-native English speakers by increasing the linguistic demands of lectures? And could intralingual live subtitling offer a form of language support that balances accessibility with academic expectations?
Questioning the conservative Flemish language policy
Flemish higher education has traditionally attached great importance to Dutch as the language of instruction. Detailed guidelines govern the use of other languages, such as English. However, this language policy is coming under increasing pressure. In the context of internationalization and globalization, the question arises as to whether the current rules are still aligned with the times. The demand for English-language instruction is increasing, especially in university programmes highlighting the importance of studying the implementation of English as the language of instruction in academic lectures—and the potential benefits of live subtitling in this context.
Investigating the effect of intralingual live subtitling
Intralingual live subtitles were added to a series of EMI lectures focusing on a variety of theoretical topics but not language courses. Each of the lectures consisted of two parts, each of around 50 minutes with a break of 10 minutes. Subtitles were provided for half of each lecture segment (around 25 minutes). The subtitles were alternately produced through respeaking and fully automatic unedited—speech recognition (ASR), but the two different production methods were never both used in the same lecture. Afterwards, semi-structured interviews and focus groups were conducted. The one-to-one interviews were held with 16 students in total, eight of whom studied English and eight of whom did not. The focus groups were organized in a similar way—a first focus group with 13 students who were studying English and a second with 9 students who were not. Comparable questions were asked in the semi-structured interviews and the focus groups. Both the interviews and the focus groups were transcribed verbatim (see research data), after which the answers from the students were analyzed for tendencies.
Results and discussion
The main findings from the qualitative data were discussed, with a focus on the preference for the subtitles production method—unedited ASR versus respeaking.
The first focus group (Focus Group 1, with English) preferred respeaking, if a single method had to suit those students studying English and those who were not. They argued that students studying English would be able to ignore the subtitles as they would not need them, while students studying languages other than English would benefit more from respeaking due to the more trustworthy information compared with unedited ASR. However, when given the choice themselves, the students from the first focus group preferred subtitles produced via unedited ASR, although they reported being comfortable with respeaking as well.
Surprisingly, the students not studying English (Focus Group 2) also much preferred subtitles produced via unedited ASR, choosing fast subtitles of lower quality over the lower projection speed of the respeaking subtitles, which they found disturbing. However, they also indicated that they did not pay much attention to the subtitles, which complicates interpretation.
When looking at the individual interviews, none of the students found the subtitles particularly useful or consciously used them intensively. This was observed in both focus groups as well. This can be explained by good English language skills and reasonably comprehensive PowerPoint slides being used for the lectures. It also related to disturbing factors, such as an excessive delay, which was mainly associated with respeaking. Delay was perceived to be distracting by 62.5% of the students who were interviewed individually:
- The delay can be useful on the one hand, but on the other hand, you hear the lecturer say something new which you cannot focus on, then you must read the subtitles again. That way, you are constantly behind. (English translation)
In the case of the subtitles produced through unedited ASR, poor quality caused the greatest annoyance. For example, 25% of the students interviewed indicated that they were annoyed by the high error rate of the subtitles produced with unedited ASR. For two students, the errors and distorted words even caused additional distraction:
- Ignoring the subtitles was difficult and I was too busy looking for mistakes and watching how the words were distorted. That sometimes gave a funny effect. My eyes automatically focused on them and ignoring them was difficult. I tried to make sure the subtitles were the last thing I looked at. So, I looked at the lecturer first, then at the info on the slide, and then at the subtitles. (English translation)
The individual interviews therefore show that although unedited ASR is a fast and inexpensive way to generate live subtitles, quality remains a stumbling block. Moreover, the students interviewed were not only annoyed by the errors but by the lack of punctuation and sentence delineation. This is why 56% of them preferred respeaking. A slight majority of the students surveyed thus considered the accuracy of the subtitles more important than the speed at which they were displayed. Interestingly, no difference could be seen here between students studying English and students not studying English. Students of English might be more bothered by the errors in the subtitles, but here they simply did not read the subtitles and therefore did not see the errors. These results are not in line with the focus group statements—both focus groups preferred unedited ASR and chose speed over quality.
Finally, the individual interviews also showed that some students demonstrated metacognitive skills. For example, some said that they listened first and then, if they had time to spare, only looked at the subtitles. Some also stated that, although they generally found the delay to be disturbing, they occasionally used the delay as a backup when they had not heard something correctly. In addition, the individual interviews revealed a link between metacognitive skills and distraction. For example, 75% of those interviewed indicated that their eyes automatically focused on the subtitles from time to time or that they looked at them out of curiosity, interest or boredom but did not perceive it as a distraction. They could ignore the subtitles if they wished. However, the remaining 25% of students failed to do so, which shows that the introduction of subtitles is not a win-win situation. During the lectures—for the students who did not need them and had difficulty ignoring them—the subtitles created a redundancy effect and more extraneous cognitive load.
Conclusion
In conclusion, why did the students not appreciate subtitles during the EMI lectures? As most students did not appear to need them, the effect of the subtitles diminished, making it hard to find out whether subtitles could be an added value in such lectures. Possible reasons for this were the students’ high language level, which sometimes made subtitles unnecessary, and the distracting factor that subtitles occasionally caused. The latter made it more difficult to follow the lecture. Accordingly, the usefulness of subtitles depends on how long the students concentrate on them: students tended to glance at the subtitles only briefly, causing them to miss the support of the subtitles and lose the focus of the lesson. Focusing longer on the subtitles is therefore key, but since this mode of instruction requires adjustment, students’’ metacognitive strategies may still be developing.
Further research into the utility of subtitles
Although the subtitles in these EMI lectures were not an unqualified success, students did recognize potential uses for subtitles in other language contexts. Some students indicated that subtitles would be desirable in lectures that focus on language acquisition—for example, in courses focused on language acquisition such as French, German or Spanish or in culturally orientated lectures delivered in a foreign language.
Yanou Van Gauwbergen is a PhD holder in Media Accessibility Studies from the University of Antwerp in Flanders, Belgium. His research focused on the effect of intralingual live subtitling on perception, performance, and cognitive load in an EMI university lecture.
