ICLC 2003

Learning and Representation of Verbal Humour:
the Case of Limericks

Greg Lessard, French Studies, Queen's University, Canada

Michael Levison, School of Computing, Queen's University, Canada

Keywords: verbal humour; natural language generation; learning

Let us begin by distinguishing verbal humour from jokes. The first is in principle non-narrative and does not necessarily provoke laughter, but rather an appreciation of cleverness. The second is in principle narrative and has as its goal to provoke laughter. (See Lessard, Levison & Venour, 2002, for a discussion of the distinction.) In what follows, we will restrict ourselves to the examination of verbal humour, leaving open the issue of whether what we find is also applicable to jokes.

There is evidence that the generation of humour is a rule-governed activity. For example, mechanisms capable of generating verbal humour may be formalised in terms of algorithms and used to generate actual humorous utterances. In earlier work, using the VINCI natural language generation environment (www.cs.queensu.ca/CompLing), we have shown this to be the case for several subclasses of verbal humour, including Tom Swifties and several classes of riddles. (See Lessard & Levison, 1992, 1993, and also Binsted & Ritchie, 1997). This does not mean that we suppose the specific algorithms used in generation to exist as internal mental representations in speakers, but only that the productions observed in humans admit of formalisation.

Inasmuch as they are formalisable, it is reasonable to ask to what extent such classes of verbal humour are also learnable. In earlier work (Lessard & Levison, 1995), we analysed an online discussion of Tom Swifties to show that over the course of the discussion, there was evidence that participants took up and extended earlier examples of puns to produce new, related, instances. In the paper proposed here, we will examine this phenomenon in more detail, using limericks as our subject matter. Limericks have the advantage of being at the boundary of verbal humour and jokes. They are highly constrained at the formal level, but they contain rudimentary narrative elements.

We will begin by using the VINCI environment to show that, like puns and riddles, limericks also admit of formalisation and generation by machine. We will then present the results of the following experiment. Adult native speakers of English are each asked to produce an instance of a limerick. These are evaluated in terms of the traditional definitions of the form (see for example Bibby, 1978). Afterwards, subjects are subjected to one of three test conditions: no instruction, provision of further examples of limericks, and explicit instruction on the form of limericks. Test conditions are randomized across subjects. Subjects are then asked to produce additional examples of limericks and these are compared with the baseline provided by their pre-test performance. The goal of the exercise is first to determine whether implicit or explicit instruction leads to learning as compared to the pre-test condition (learning being measured as improved performance), and second whether there exists any measurable difference between the results of implicit and explicit instruction. Discussion of the results will be used as the basis for more abstract consideration of the mechanisms of humour from the point of view of production.


Bibby, Harold Cyril. (1978) The Art of the Limerick. Hamden, Conn. : Archon Books.

Binsted, Kim; Ritchie, Graeme. (1997) Computational rules for punning riddles. Humor, 10 (1), pp.25-76.

Lessard, Greg; Levison, Michael; Venour, Chris. (2002) Cleverness versus funniness. April Fool's Day Workshop on Computational Humour, Trento.

Lessard, Greg; Levison, Michael. (1997) Rule-governed wordplay and creativity. In Mind II: Computational Models of Creative Cognition, Dublic City University. http://www.compapp.dcu.ie/~tonyv/MIND/greg.html

Lessard, Greg; Levison, Michael. (1995) Linguistic and Cognitive Underpinnings of Verbal Humour. International Cognitive Linguistics Association Conference, Albuquerque.

Lessard, Greg; Levison, Michael. (1993) Computational Modelling of Riddling Strategies. ACH/ALLC Joint Annual Conference, Georgetown University, Washington, DC. (Extended abstract in conference proceedings, pp. 120-122.)

Lessard, Greg; Levison, Michael. (1992) Computational Modelling of Linguistic Humour: Tom Swifties. ALLC/ACH Joint Annual Conference, Christ Church, Oxford. (Extended abstract in conference proceedings, pp. 175-178.)