I MPLICATIONS FOR TEACHING

The various relationships between language and the learning of data modelling that are described in this thesis have consequences for the planning of teaching sequences.

Each of the four research papers offers a section on suggestions for teaching. In this section, I will summarise these points and also provide a few additional comments.

The students need help in the process of realizing the different semiotic systems or

language games involved when they engage in data modelling as a social activity.

Even though experts do not seem to take notice of these distinctions (Dreyfus &

Dreyfus, 1986), it has been demonstrated through the research presented in this thesis that novice students need to be more explicit in the way that they handle the various terms and expressions involved when solving a data modelling problem. They need to have metalinguistic awareness.

The distinction between natural and artificial languages play a particularly important role in this respect, both for scientific concept building and for labelling of attributes and entities. It is therefore advisable that the teacher spends some time focusing on different types of language games, and on the meaning of utterances as defined through actual use, and not from predefined rules or by definition in a dictionary. Teaching scientific concepts to novices, it is furthermore important not only to explain how the constructs are intended to function, and what their meanings are (from an established scientific point of view). One needs also to emphasise how these understandings of the constructs differ from the natural language use of the corresponding terms. In science teaching this has been addressed by sharing with students the idea that learning science (or computing) involves learning a new language to talk and think about familiar things. In some cases the terms are familiar and carry familiar meanings; in other cases the reverse is true (Mortimer & Scott, 2003). Paper 2 offers some further explicit suggestions of possible classroom activities that may be adopted in order to improve the metalinguistic awareness of the students. These are mainly concerned with helping the students realise the distinctions.

The next step is to provide situations where the students practice formulating explicitly what they mean by the terms they choose to use as labels, and to distinguish between this meaning and their pre-knowledge from everyday language. In the related field of mathematics education, Shoenfeld (1992) describes positive results from an experiment trying to improve students’ problem solving capabilities through increased metaknowledge. In the experiment, the teacher continuously moves around the classroom stopping at the table of individuals or groups and always asks the same set of questions; i.e. “ What (exactly) are you doing?”, “why are you doing it?” and “How does it help you?” (Shoenfeld, 1992: p356). Eventually the students got used to the questions, and were generally better prepared to give a satisfactory answer when the

teacher approached them. After a while, the teacher was able to cease the questioning as the students were now asking themselves the same set of questions each time they came to a decision point. This approach can easily be adapted to a computer science classroom, and to focus on the metalinguistic awareness of the students instead of their problem solving capabilities. Questions that could be used include: “What have you labelled that entity?”, “What exactly is the meaning of the entity?” and “How does that meaning correspond to the term you have chosen?”. The answers to these questions would address both the issue of labelling and use of vernacular lexis, and the issue of creating abstract entities for relational phenomena.

The discussions presented in this thesis deal with various language systems that operate on different levels in parallel. In addition to the vernacular–technical dimension discussed in connection with everyday terms used as labels (i.e. the semiotic dimension), there is the metalanguage hierarchy of data modelling defined as a semi-graphical language for making formalized descriptions of the problem domain.

It will be unnecessarily complex to introduce all of these dimensions to the students, but some of these issues should be addressed. “Learning technical discourse implies learning the lexico-grammatical language of that discourse, which, for science, implies learning to transform everyday or vernacular language into an uncommon-sense language.” (paper 4: p17). Considering the semantic framework introduced in paper 4, a set of questions that could be helpful for increased metalinguistic awareness are: “What do you mean when you use that term?” and “In which contextual frame do you understand the term when you use it like that?”.

Another significant finding of this research is the local development of language games in each particular classroom. In order to facilitate this development, it is imperative that the students are allowed to interact discursively with each other to a sufficient extent. The teacher needs to be alert and able to adapt the way he or she uses scientific language in interaction with the students. The teacher’s responsibility in this sense is twofold. First, he or she offers an invaluable benchmark or point of reference for the students as they develop their own version of the scientific discourse.

Second, and equally important, the teacher needs to adapt to the discourse of the classroom as it evolves. In this way, the teacher will function as a participant in the discursive social practice and thus be able to influence the development toward fruitful viable ways of doing computer science with language. It can be advisable to

even focus explicitly on this aspect of the discourse and compare the conceptual networks developed in the student group to the established text book scientific discourse. This is similar to what Lemke (1990) did in his studies of the development and use of thematic patterns in the science classroom.