3 Setting and Methodology
3.3 Methodology
3.3.3 De-limiting the context
In acknowledging the importance of context in shaping the actions and utterances of human beings, the researcher is left with a problem: that of attempting to de-limit the seemingly infinite number of contextual factors that could be taken into account, even in the restricted environment of the
classroom. In aiming to do this, a useful starting point is to acknowledge the centrality of talk to my research concerns. Indeed, talk is seen here as context-shaping, transforming the setting and the social relations within it (Bauman and Briggs, 1990: 68). From such a viewpoint, context becomes a fluid concept, something which it is within the powers of the human agents
22
I usually played back the recordings the day after making them and, in the case study, I recorded the material in the first half of the week and played it back in the second half, taking notes during playback while the interactions were still fresh in my memory.
96
present to shape and change. Furthermore, the verbal and non-verbal actions which take place are highly complex and interrelated. Attempting to separate them out as component parts may of itself distort them. However, some form of selection must take place to make the data manageable. In doing so, a simple fact should firstly be acknowledged which sometimes goes unmentioned in social research: investigators are themselves
experienced social actors. As such, their noticing of certain phenomena in the stream of experience is itself significant. In investigating something like HLP in everyday interactions, the analyst necessarily has to be selective in deciding what is noteworthy. Even when asking participants for their own perspectives on recorded interactions, it is the analyst who, for practical purposes, usually pre-selects those exchanges which are of significance. This is especially true in the field of HLP where instances of such play may not occur for long stretches of natural interaction. As Holmes (2000: 163) points out in explaining the research methodology she employs in her own research:
The analyst’s identification of instances of humour is a crucial component in the analytical process [...] Instances of humour in this analysis are utterances which are identified by the analyst, on the basis of paralinguistic, prosodic and discoursal clues.
Of course, language is both the medium and subject matter of instruction in the classroom and the main means for accomplishing play, even for those learners who have little proficiency in the TL. One means of identifying
relevant data in this investigation is to look for the various surface features of language which seem to occur during language play episodes. Goffman
97
(1974: 42) observes that there is a great deal of repetitiveness in a play frame; Coates (2007: 43) identifies syntactic repetition while Carter (2004: 89-112) notes the use of neologisms and tropes more commonly associated with literary texts. Research in the field of SLA echoes some of the findings from investigations into NS interactions. Broner and Tarone (2001: 371) note unusual lexical choices such as neologisms while other researchers (Bell, 2005; Belz and Reinhardt, 2004) demonstrate how proficient L2 speakers are able to playfully exploit the forms of the language.
However, looking for such surface features would necessarily preclude those learners (probably the majority) who lack the expertise for a native-like
exploitation of forms, yet are nevertheless able to play in the TL.
Furthermore, a more fundamental issue arises when researching social interactions. As Jones (2007: 341) puts it:
Words do not produce or interpret themselves; people, engaged over some matter, are responsible for that, and, under certain
circumstances, answerable too.
In other words, looking at the surface forms of language is not enough. Play can only be so classified if the participants in it regard it as such. The
features of utterances do not alone constitute evidence of anything. It could be argued that this is especially true when investigating HLP. Not only does the deniability of much playful language use allow something to be said but not said (North, 2007: 553), but ‘humour often lies in the gap between what is said and what is meant’ (Coates, 2007: 32). People derive pleasure from
98
understanding each other without having to articulate everything being communicated.
This leaves a problem. It means that an ‘etic’ classification of linguistic features i.e. one based on criteria formulated outside the group, is, on its own, inadequate. However, an 'emic' perspective (Cook, 2000: 67) - one formulated by the participants themselves - on which episodes constitute play was, in my research context, practically impossible, given the time constraints which course participants are under and other obstacles (see Section 3.3.5 for a discussion of why learners were not interviewed). However, one means of identifying whether the participants regarded particular exchanges as playful was by pinpointing moments of laughter. Indeed, unsurprisingly, this is a method used by a number of researchers in this area to find significant episodes of play, both among NSs (e.g. Carter, 2006) and NNSs (Bell, 2005: 198; Cekaite and Aronsson, 2005: 174). Although laughter is a useful pointer to play, it is, of course, by no means foolproof as a detection device as it may, for example, merely signal support or something less benign such as embarrassment or even anger.
Fortunately, it has already been noted (Section 2.4) that play, being a stepping away from the norm, is usually carefully signalled by its instigator through contextualisation cues (Goffman, 1974: 45). Marked or contrasting prosody can be good indicators of such episodes (Bell, 2005: 199; Broner and Tarone, 2001: 363; Pomerantz and Bell, 2007: 563) as can smiling (Bell, 2007a: 39). An expansive and exaggerated body language may also indicate that a play frame is operative (DaSilva Iddings and McCafferty, 2007: 42). Indeed, intuition suggests that the latter would be a very useful resource for
99
learners needing means other than language to signal play. A further indication of play is that it can affect the structure of talk, disrupting the
normative features of a particular type of classroom activity. For instance, the learner may 'topicalise' (van Lier, 1988: 152) - assume control of the topic of communication - at a moment where, ordinarily, he or she would not have the right to do so. Another aspect which facilitates the identification of play is the fact that it has a tendency to occur in clusters (Carter, 1999: 199-200; Holmes, 2007: 530; Norrick, 1993: 42). This seems to arise from an impulse to join in the ‘fun’ and as a signal of camaraderie. Furthermore, Goffman (1974: 43) posits that, in humans as in animals, the openings and closings of play frames are usually cued clearly in order to avoid misunderstandings. The research cited above indicates the importance of contextualisation cues in signalling when people are entering and then participating in a play frame. Indeed, what research there has been into play in a second language
indicates that participants can identify and appreciate such cues, even when the accompanying humour is either not, or only partially, understood (Bell, 2007b: 377).
In sum, although the researcher has to be wary of risks in identifying play - contextualisation cues do not necessarily carry the same meanings across cultures (Gumperz, 1982) and body language especially is open to a great degree of interpretation (Adolphs and Carter, 2007: 136) - there are strong indicators which can help him or her in the task. Play episodes are usually cued by the play instigator. Their initiation and development can be
recognised also through the reactions of the others present and evidence of changes in the structure of talk. In addition, play tends to cluster. It is,
100
perhaps, less easy to identify the ending of a play episode, precisely because of the clustering that can take place. However, signals, verbal or otherwise, that indicate a return to the primary activity can usually be
discerned. Also the structure of talk may well return to the typical patterns of interaction that were operative before play disrupted them. In other words, 'normal service' can be seen to be resumed.
Another criterion for selecting particular episodes for analysis needs to be mentioned. At the outset of this thesis, I explained how my interest in investigating HLP among my language learners was triggered in part by noticing their laughter inducing in-group references. The event or exchange from which particular references grow need not necessarily be particularly playful in its origin. Therefore, in tracing the development of incremental play, some exchanges only gain significance with hindsight. Therefore, the
selection process is, in part, a retrospective one. In this regard, it should be mentioned that, in the investigative process, certain patterns begin to emerge such that an exchange can become more significant in the light of previous exchanges noted in other groups and at other times.
Group laughter and clustered stretches of joint banter and fun with the language indicate a successful play frame. However, the current research also finds interest in those moments where an attempt at humour fails. In exploring how learners are able to have fun with language despite a lack of common socio-cultural reference points and an incomplete control of the language, it is instructive to investigate failures as well as successes (Bell, 2007b). Surface features to pinpoint such failures are not always easy to find. However, discordances, asymmetries and arrhythmic exchanges often
101
signal communicative breakdowns of this sort. It has to be acknowledged, however, that by their nature, some failures in humorous play may well go unnoticed.
Given the difficulties set out above in identifying pertinent segments of data, measures are needed to validate the process of pinpointing, classifying and analysing relevant material for this research. These will be set out later in this chapter. Firstly, the means of collecting data will be explained.