2. Description of the project and analytical approach
2. Description of the project and analytical approachThe data comes from a corpus collected in 2001-2002 in the context of a project called Tic-Talk, after 'Technologies de l'Information et de la Communication' in French for 'ICT' in English. The primary aim of the project was to encourage intercultural understanding between anglophone and francophone secondary school pupils by providing them with opportunities to communicate directly with pupils who were speakers of the language they were learning and who were learning their own language. Key features were that the pupils:
Tic-Talk was not conceived as a research project nor as a teaching programme but as a voluntary activity which the pupils could carry out independently on their own in a secure Web-based environment. It was aimed at the classes which did not have public examinations and were less pressurised. The pupils' ages ranged from 14 to 18 . There was no compulsion, marks or incentives and pupils could participate or opt-out at any time. The 6 teachers and 3 lecturers involved did not intervene once initial instructions had been given. The teachers' role was to select pupils and to send their names to the WebBoard manager in the Faculty, to facilitate pupils' access to WebBoard if needed and to observe what happening.
The project was initiated by staff in the Faculty of Education of the University of Cambridge and a network of modern language teachers to explore the use of new technologies for exchanges between pupils. A pilot had taken place the preceding year. Six schools participated, three of which were in South East England. The French-speaking schools were in Belgium, in France, and in Senegal. The ideal Tic-Talk group was originally intended to have 6 participants: three anglophones and three francophones. There was little pupil response in the UK so that eventually most groups turned out to have only four 'sites', or e-mail addresses, one for each country involved. The number of email addresses and the number of individual pupils did not always.coincide. In Senegal, because of the large number of pupils and the number of computers, pupils were organised in small teams with a group secretary so that the Senegalese 'address' could represent up to four pupils.
The programme lasted 4 months from October 2001. It was interrupted several times by holidays whose dates did not coincide in the different countries so that groups functioned for a maximum of eight to ten weeks. The 24 groups involving 152 pupils functioned independently from each other but locally, even pupils who were not active participants could monitor the groups' exchanges. All participants granted permission to have their messages analysed for research purposes but all names have been changed in the examples to protect their identities, the rest of the examples being otherwise strictly identical to the original messages. The present paper is based on the analysis of 12 groups chosen randomly and representing 75 pupils. They are distributed over 53 addresses as follows: Belgium: 15, Senegal: 12 representing 34 pupils, France: 13 and UK: 13.
In view of the aim of exploring WebBoard as a new arena of language use, we wanted to triangulate the methods of analysis of the discourse data not only to improve the validity of the study but because it seemed the best way to approach the exploration of a complex phenomenon.
(a) We first look at collaborative discourse from the point of view of the effect of collaboration on the patterns of language use in the data. We take the view that the pupils form a kind of bilingual group and that on the assumption that the participants are trying to use their L1 and L2 as suggested, which would predict a 50-50 split between the two languages overall, patterns of language choice in the discourse reflect the degree to which they are working together. This amounts to taking a naïve 'common sense' view of collaboration (literally meaning 'work together'). Basic arithmetic is sufficient here to characterise the overall corpus in terms of the amount of joint work carried out, as measured by the distribution of English and French in the pupils' messages. This crude analysis provides us with a snapshot representation of the overall degree of collaboration between the pupils which we can teeze out in the following qualitative analysis.
(b) The view of collaborative discourse we adopt for the main analysis comes from Herbert Clark who proposes that language use is a form of joint action taken by participants who share information (Clark, 1996, p. 52-56). A conversation is a co-ordination game in which the shared common ground is constantly updated into a 'discourse representation' composed of a textual representation, 'a representation of the language and other signals used during the discourse' and a situational representation, 'a representation of the situation being talked about.' These can be traced in the 'discourse record', or the text of the actual messages in our data. It represents the official or 'on record' states and events in the current activity. Public states and events that are 'off-record' are not official parts of the joint activity. Clark's model has distinct advantages for us as analysts:
The concept of signalling as co-ordination device in discourse offers descriptive tools which are neutral as regards to conventional methods of signalling such as the use of linguistic symbols as in 'English' or 'French' (Clark 1996, p.188). This means that cross-linguistic discourse cohesion whereby an English noun, say 'chair', is referred to as 'elle', a feminine pronoun, in the following French message is not problematic. Also, participants may use non-linguistic methods to convey their intentions such as gestures which can demonstrate for example, how a person holds an umbrella, or such as gaze which can indicate who one is referring to. Although co-ordinating devices have been studied in face to face interaction, the possibility of creating new symbols and icons is open whenever joint activities are taking place. Moreover, in WebBoard, pictures and diagrammes can be posted.
Viewing discourse as a joint activity means that there is a discourse continuum on which joint activities can be positioned depending on how heavily they rely on conventional language. A telephone conversation is 'mostly linguistic' but a tennis match 'least linguistic'. We thus have a measure to place the activity ' WebBoard conferencing' on the continuum.
The model lends itself well to the analysis of asynchronous text-based communication where the interactional order is not problematic, in sharp contrast with chatroom discourse, and in which the sequence of messages makes it possible to make a message by message analysis of the discourse record as interpreted by the participants in the context of use.
Finally, the model does not undermine the role of discourse as social practice and socio-cultural meanings in interaction, such as demonstrated by interactional sociolinguistics and especially Gumperz 's work (1982). In particular, it makes it easy to include signals which have no other intention than to re-assert that participants share a common view point.
The model was used to develop a crude evaluation measure of whether messages contributed to the discourse: each message was coded independently of language choice as Fail (F), Co-operative ( C ) or Collaborative ( C*). F meant the message did not contribute to the discourse i.e. there was no sign of any kind that the message had been acknowledged by anybody in the discourse record.( C ) meant the message contributed to the discourse but that it was 'minimal' in the specific - sense that its absence would have been uncooperative and would have required some sort of an explanation. A typical case would be the answer to a question sent to a specified addressee (C*) meant the contribution initiated a new exchange - thus displaying an intention to increase common ground further- or called for further work, for example, asking for some sort of clarification.
