Description

Introduction
One of the effects of the Internet's world-wide explosion in popularity has been the increased emphasis on the international use of English. Discussion lists which span cultures have, in most cases, adopted English as their official language, since it already enjoys wide use among professionals and academics.

The popularity of some discussion lists is attested to by their large numbers of subscribers. Yet, in many cases, the actual number of active contributors is relatively small. Among the many possible reasons for this is one which has great significance for education and personal development in Brazil: the timidity of list subscribers who lack writing proficiency.

Justification
There are list subscribers who would contribute and, thus, derive benefit from the interaction if they had either, more confidence, more proficiency, or both. As a group they exhibit two characteristics which suit them for computer mediated distance learning (CMDL): they are all experienced computer users and they are dispersed over a large area. CMDL offers an opportunity for these individuals to defeat the constraints of geographic isolation and participate more fully in developing themselves and the nation.

From a pragmatic point of view, the most direct way of offering assistance would be a CMDL course based on direct observation of how language is used (Austin 1962; Searle 1969; Halliday & Hassan 1985). Hutchinson & Waters (1987) accent the importance of observation by the vital role they give to needs analysis in developing ESP courses. In the present case, the language being examined and analyzed is the interaction carried on in e-mail messages. These messages are the essence of discussion lists and, as such, provide the key to aiding the further development of Internet discussion list participants in Brazil.

Aim
Since discussion lists are, in most cases, public, a corpus of e-mail messages was readily available. Therefore, the collection and analysis of e-mail messages was carried out to furnish a basis for a CMDL course in written English. The ultimate objective of this project is to increase the efficiency and enjoyment of discussion list interaction for those who would benefit most immediately and that the knowledge they gain will be diffused throughout their communities.

Method
On the basis of their popularity and importance within their respective areas, three Internet discussion lists were chosen as data sources: TESL-L (English teachers and education researchers), LINGUIST (linguistics and language research) and CHEMISTRY (the use of computers in chemistry). They are all lists which use English as their official language and generate 50+ messages per week, on average.

Groups of messages from the same period were downloaded from the public archives of old messages maintained by all of the lists. Then, from each group, forty messages were randomly chosen for analysis. The next step was to define the instrument of analysis to be used in identifying and quantifying characteristic language in the final corpus of 120 messages.

The development of the analytical tool began with Eggins (1994) by way of Halliday (1985). In her chapter, "Systems: meaning as choice", Eggins outlines a "Speech function system (discourse-semantic stratum)". The system, as the author describes it, is a "formalism used in the systemic model to capture paradigmatic relations...."(:205). Moreover, it consists of eight inter-related choices, some of which are dependent on previous choices for their implementation:

1) giving/demanding
2) goods & services/information
3) initiating/responding
4) subjective orientation/addressee orientation
5) inclusive/exclusive
6) querying/questioning
7) fact/opinion
8) supporting/confronting

Since some categories are dependent on others, some choices proved to be more fundamental and general than others. The first three choices are of this type. In fact, they are so fundamental that they were applied to whole messages first. Afterward, each message was analyzed sentence-by-sentence, using all eight choice groups. The resultant mixtures of sentence types were then looked at in relation to their respective message types to establish a relationship between sentence types and overall message function. Analysis results were used in determining the content of a proposed CMDL course in written English. Data derived from the analysis permitted the identification of language commonly used in the discussion list environment. Subsequently, this data was incorporated into the syllabus.

Results
Preliminary evaluation of the results revealed two trends:
1) the great majority of messages were in response to ongoing discussions;
2) all messages exhibit a surprising variety of sentence types.

The preponderance of responding messages makes sense in light of the ease with which users can quote and reply to e-mail. One question or request can trigger a veritable avalanche of responses. As a result greater attention was given to the necessity of teaching language functions used in offering information. Second in importance is language used in the various manners of questioning or querying. By acquiring competence in these functions, discussion list participants can interact with more effectiveness and confidence.

As to the mixture of sentence types, it was noted that in several cases the nature of request messages could be determined by a simple question at the end of an extensive passage of information or explanation. These messages often took the form: "I did 'A'. I did 'B'. I did 'C'. What do I do now?" It was further noted that the relationship between the question and the information could be inverted without changing the function of the message. A determining factor here seems to be whether the message initiates the interaction or is in response to someone else's initiative. Some of the questions in responding messages seem to be turn markers which indicate that the writer doesn't think that the topic is exhausted or that his/her text is final.

Another aspect of the mixture was the location of specific language within larger structures (paragraph and text). The results show, among other things, a variety of possibilities for beginning and ending requests and responses. Analysis of the data demonstrates where and when specific language is used. For instance the use of "does anyone know..." at the beginning of a text can be compared to and offered as an alternative to using "can you help?" at the end. How these two forms are used can be demonstrated in contexts derived from the corpus. Some message types were conspicuously absent due to their nature and that of the discussion lists. Messages concerning the exchange of goods and services are generally prohibited since they preclude financial transactions.

Semantic analysis of the type proposed in this paper is a useful tool in ESP type needs analysis and can be readily applied to CMDL situations in Brazil. By assuring that distance learners receive exactly what they need CMDL better serves the community and society at large.