ELFA project

Getting serious about laughter in academic talk


Academic discourse is serious business. Lectures are delivered, conference presentations are discussed, great thoughts hang in the air like disembodied spirits. It’s not the kind of environment you’d expect to find a lot of laughter and joking. And yet, we academics can’t seem to stop laughing.

The Baltic Sea is still frozen in February. We’re anxiously awaiting the sun.
© Nina Valtavirta

The ELFA project had our February meeting on Thu., 21.2, and MA student Jani Ahtiainen gave a talk on laughter in spoken academic discourse. He’s doing his master’s research on terms of address in the ELFA corpus, an area often connected to culture-specific norms and expectations. Likewise, the occurrence of humor and laughter might be influenced by culture as well.

Jani based his discussion on a 2006 article by David Lee that looked at occurrences of laughter in MICASE (Michigan Corpus of Academic Spoken English). The idea behind the article is that foreign students must struggle with the profound subtlety of American humor, so we should study laughter in MICASE to help these hapless foreigners cope. These are quite different research motivations than we have in the ELF field, but the question of laughter in academic ELF is still relevant.

Humor in a foreign language

Some of the group discussion during Jani’s presentation dealt with perceptions of humor in a foreign language. There seems to be a folk linguistic belief that one can’t be funny in a foreign language (though of course this varies between people). As anecdotal evidence, Jani reported that it was difficult to find examples in ELFA of deliberate “jokes”. So is it safe to assume that with intercultural, lingua franca interaction in the sophisticated world of international academia, we’ll find dramatically less laughter in ELFA than in MICASE?

The answer is no. ELF users evidence nearly as much humor as their academic counterparts in the States. Jani followed Lee’s basic approach by counting the number of laughter tags in ELFA and checking their distribution across the corpus and types of speech event. The raw counts of laughter tags from the two corpora are regularised based on their size: MICASE is 1.8 million words covering 195 hours of speech, and ELFA is 1 million words covering 131 hours. This is how they compare:

corpus laugh tags (raw frequency) laughs per min. laughs per 1,000 words
MICASE 5542 1 laugh per 2 min. 3 laughs per 1K words
ELFA 2193 1 laugh per 3.5 min. 2 laughs per 1K words

While the frequency of laughter is higher in MICASE, the difference between the corpora isn’t as great as might be expected. Moreover, the frequency per 1,000 words is likely the better gauge, as the ELF talk naturally progresses more slowly than that of native speakers: there are about 150 words per minute in MICASE, with less than 130 words per minute in ELFA. It thus appears that laughter occurs at a quite similar frequency in ELF and native English academic discourse.

What’s so funny?

When laughter is examined in different types of speech event, Jani’s findings also corresponded to Lee’s. The more interactive event types yielded more laughter in both corpora. In ELFA, the speech events with at least two laughs per 1,000 words were seminar discussions, conference discussions, and lecture discussions. Among the 20 events with the most occurrences of laughter (all multi-party discussions), all had over 4 laughs per 1,000 words. The highest-ranking event – a medical seminar discussing the hilarious world of “abdominal complaints” – clocked in with a whopping 14 laughs per 1,000 words. Good times!

So, clearly the world of academic ELF isn’t as joyless as one might think. Future research could fruitfully focus on who is laughing, and why? Having personally converted the ELFA corpus to XML format – which involved manually checking the synchronisation of all overlapping speech in the corpus – I have anecdotal impressions of my own.

I would hypothesise that laughter in academic ELF often corresponds to signalling linguistic markedness (something that is unusual or unexpected). It’s common to find laughter co-occurring with idiomatic expressions, words coined for a specific occasion, as well as “big words” (aka “five dollar words” in the States), as if to signal that a linguistic item deserves some special attention. Though the frequency of laughter in ELF may be similar to native English, I’d expect to find different functions in ELF talk.

Research openings with XML

Lee’s research is based on searching for the XML tag for laughter using WordSmith tools, a popular corpus analysis application. But this approach hardly draws on the power of XML, which, as computer-readable structured data, can be queried using a programming language such as Python. It’s not profoundly difficult to write a program that not only counts all instances of laughter, but also associates with it all the speaker metadata (gender, age, role, first language) and event data (type of event, academic domain) and export it to a Comma Separated Values file for further analysis in a spreadsheet application.

In later months, after I’ve dug myself out of my latest slew of urgent deadlines, I’d like to write such a program for the ELFA XML corpus and post some original findings here. Maybe an MA student would like to investigate the data further. I find interesting stuff all the time that I don’t have time to investigate in depth, so I hope this blog will also be chance for others to get ideas for their own research.


Lee, David (2006). Humor in spoken academic discourse. NUCB journal of language culture and communication, 8 (1), 49-68. Available online: http://ci.nii.ac.jp/naid/110004860604/en (23.2.2013)

ELFA 2008. The Corpus of English as a Lingua Franca in Academic Settings. Director: Anna Mauranen. http://www.helsinki.fi/elfa/elfacorpus. (23.2.2013).