3.4 Modes of Discourse and Conversational Frameworks

Sinclair’s and Coulthard’s seminal work (1975) might be investigated as the basis of analyzing the discourse modes in text-messaging.  In analyzing classroom discourse, they devised a hierarchical structure comprising acts (elicitations, directives and informatives), which in turn structure moves (framing, focusing, opening, answering and follow-up), which in turn realize exchanges.  Examples of the latter include ‘teacher inform’, ‘teacher direct’, ‘teacher elicit’, ‘pupil elicit’, ‘pupil inform’, ‘check’ and ‘re-initiation’.

However, Sinclair and Coulthard themselves cautioned against the broad applicability of their scheme to contexts outside of traditional classroom talk.  In fact, as Edwards and Westgate (1994) point out, Sinclair and Coulthard were not educationists, but linguists.  Their work was therefore not primarily intended to inform strategies of teaching and learning.  In contrast, Cazden (1988) chose to apply herself to the particular context in which it is not necessarily the case that “one participant has responsibility for the direction of the discourse, for deciding who shall speak when, and for introducing topics” (Sinclair and Coulthard, 1975: 6) – that is, to “the only context in which children can reverse interactional roles with the same intellectual content, giving directions as well as following them, and asking questions as well as answering them [emphasis added]” (Cazden, 1988: 134), namely, pupil-pupil discourse.

In the present study, the analysis of the messages which the pupils sent to each other during their participation in the field-based interventions was developed retrospectively, using the constant comparative method.  Because the messages were sent from mobile phones, they were automatically time-coded.  This afforded easy transfer to a tabular presentational format of the messaging exchanges, in which each message was taken as a single turn. 

The analysis falls within what Edwards and Westgate term the linguistic-pedagogic approach to discourse analysis, in that the transcripts were analysed with respect to how the participants went about “talking to learn” (Edwards and Westgate, 1994: 152) – that is to say, how the non-co-located pupils sought to teach and learn from each other geographically meaningful elements within, and issues concerning, their local environment.  This focus on the co-construction of such tertiary artifacts (Wartofsky, 1979) was deliberate, because as Hughes and Westgate (1990: 47) suggest, “activities based on shared imagination have a particular potential for practice and development of communicative skills”.

There have been a variety of views with regards the most appropriate unit of analysis with respect to the general field of electronic messages, online conferencing, and e-learning.  For example, Garrison, Anderson, and Archer (2001) have used a message-level analysis, while Fahy (2002) has analysed at the level of the sentence.

In the present study, the unit of analysis, with respect to the messaging exchanges, was the individual sentences / clauses detected in each of the messages (to the extent that these existed in the first place, given the informal nature of the discourse of text-messaging amongst adolescents).  This decision was taken in order to circumvent the fact that some of the text-messages were rather long, and comprised a few ideas concatenated into a single message (for example, “Where are u now? We are saying blk 525 first.. Then find e mrt track.. U noe where izit?”).

Similar difficulties were also encountered by Garrison, Anderson, and Archer, who worked around this issue by adding more ‘clarifying’ descriptors to their schema.  In fact, the latter attributed the “problematic” results obtained by Garrison, Anderson, and Archer to their use of message-level analysis (Fahy, 2002: 10).  Fahy further justified his choice of sentence-level analysis by suggesting that computer-mediated collaborative postings contain both social- and task-related material, and that, ultimately, “the two dissimilar approaches to transcript analysis arrived at a similar picture of the resulting content and processes occurring within the conference” (Fahy, 2002: 24).

More recently, Gee (1999) has written extensively on the role of discourse both as a characteristic of, and in building, cultural identities and social groups.  He has declared that the twofold primary functions of human language are to scaffold the performance of social activities and to scaffold human affiliation within cultures, groups and institutions.  Given such a premise, it should come as no surprise that the theory of discourse which he subsequently elaborates upon is founded upon investigations into six so-called ‘building tasks’.

These tasks are defined as what we do with language to build a network of communicative social interaction, and comprise semiotic building, world building, activity building, identity- and relationship-building, political building and connected building.  Taken together, these tasks serve as a plumbline against which to evaluate the validity of any discourse analysis.  Nevertheless, Gee himself acknowledges the limitations of force-fitting his tools of discourse analysis to research on new media, “anticipating that these tools will be transformed, or even abandoned, as readers invent their own versions of them”.

Because of the proprietary nature of the discourse analysis employed in the present intervention, it is necessary to consider the extent to which it is valid.  Gee (1999) suggests the following four elements of the validity of any discourse analysis.  First, discourse analyses tend to be more valid and trustworthy the more they offer convincing answers to questions raised by a consideration of his six ‘building tasks’.  To take the present study for instance, semiotic building tasks are addressed by the examination of the photographs taken by the students in their attempts to guide each other along the orienteering routes, world building tasks are addressed by the deduction of the situated meanings attached to various landmarks in the neighbourhoods, activity building by situating the orienteering activity within a larger framework of an investigation into spatial perception, relationship building by assigning distinct roles to each participant during the field study, political building by presenting participants with the need to construct and share social capital, and connection building by deducing more abstract categories of discourse from those originally identified.

The second element which Gee suggests sheds light on the validity of discourse analysis are the extent to which ‘native speakers’ of the social language pertinent to the intervention agree that the analysis reflects how the language actually functions.  As elaborated upon below, under Lincoln’s and Guba’s (1985) criteria for trustworthiness, such member checks were indeed carried out and they corroborated the analysis.  Gee’s third element of validity is the extent to which the analysis can be applied to related sorts of data.  Investigating this is beyond the scope of the present study, and could well form the basis of a subsequent research endeavour.  Finally, Gee suggests that discourse analysis would be made more valid if it is tightly tied to the grammar of the social language under scrutiny.  In the present study, this was ensured by using the constant comparative method to ensure that the discourse modes recorded through the text-messages were thoroughly and accurately represented.

To summarize the preceding discussion, this data collected for this thesis has driven the design of a proprietary taxonomy of discourse types, associated specifically with the synchronous exchange of electronic messages in peer-based learning environments in which the collaborators are not necessarily co-located.  This proprietary taxonomy has been designed against Gee’s building tasks to ensure its validity.  The taxonomy uses, as its unit of analysis, a sentence- / clause-level descriptor within each electronic message. 

Table 2 presents the results of a hierarchical cluster analysis conducted on all the various discourse types used by participants who were members of ‘leading’ pairs during the orienteering task.  As stated above, ‘leading’ pairs were those who had been trying to communicate the given route to their peers.

From the table above, it was decided to divide the discourse types into three groups as follows:

Task frames
•    Questions checking for understanding (QCUL)
•    Questions seeking to determine or verify attainment, as well as those requesting reports on attainment (ATTAINL)
•    Questions or statements addressing misconceptions (MISCONL)
•    Instructions and re-instructions (INSTRUCTL)
•    Exploratory questions (EXPQNL)

Interactional scaffolds
•    Questions seeking elaboration (QSEL)
•    Conditional presentations of alternative (CONDALTL)
•    Directed tips (DIRTIPSL)

Acknowledgments
•    Declarations of lack of understanding (LKUSTDL)
•    Affirmations (both positive and negative) (AFFIRML)
•    Assurances and edifications (ASEDIFL)

Table 3 presents the results of a hierarchical cluster analysis conducted on all the various discourse types used by participants who were members of following pairs during the orienteering task.

From the table above, it was decided to divide the discourse types into three groups as follows:

Clarifications
•    Unsubstantiated guesses (UNSUBF)
•    Declarations of lack of attainment (LKATANF)
•    Declarations of lack of understanding (LKUSTDF)
•    Statements expressing urgency (URGENCF)
•    Statements or requests raising alert to teacher’s implicit assumptions (including declarations of non-congruence) (ALERRTF)

Follow-ups
•    Demands for information (DDSINFOF)
•    ‘Helicopter’ imperative (HELIF)

Task frames
•    Questions seeking elaboration (QSEF)
•    Questions seeking purpose (PURPOSEF)
•    Exploratory and procedural questions (EXPROCF)
•    Statements or questions seeking to either verify or report attainment (ATTAINF)

<- 3.3 Evaluation of design          -> 3.4.1 Task frames (leading pair)