I have earlier been teaching Sámi and working with vitalization of the Sámi language for about 15 years, and therefore it has been natural to look at the possibilities for using language technology as a tool for pedagogical applications. Together with my colleges I made a North Sámi version of VISL-games, developed at Syd-Dansk University. Links for Sámi VISL and documentation can be found at OAHPA! website.

The last years I have been working together with my colleagues from the Giellatekno-group on a project, which uses a morphological analyser/generator as the basis for grammatical exercises for students learning Sámi. The second part consisted in making interactive dialogues, where the machine asks the student questions, analyses the answers grammatically, and on the basis of this analysis, gives appropriate feedback to the student. Both, the dialogue progress (choice of continuation path) and the pedagogical feedback (comments on errors in the input, e.g. case errors) are based upon the analysis of the student's response. The six programs (Leksa, Morfa-S, Morfa-C, Numra, Vasta and Sahka) for North Sámi are presented under the name OAHPA! at our OAHPA! website with links to the programs. I have also been working with Oahpa for South Sámi, consisting 4 of the programs.

In the dialogues, the set of possible answers is restricted. In my dissertation project I will work on less restricted dialogues and go beyond dialogues. The goal is to recognize and diagnose language errors in any text written by students learning Sámi. I will also investigate to what extent it is possible to give pedagogical feedback.

A student learning Sámi will do other kind of language errors than a first language user, and the errors will be dependent on the student´s first language. The student will probably make complex errors which can be difficult for the analysator to diagnose.

My methodological basis for detecting and diagnosing errors will be finite state transducers for the morphology, and constraint grammar for disambiguation and syntactic analysis and error detection. One of the main challenges for my approach consists in the work with a position-based system for a language with a free word order than in e.g. Norwegian and Swedish. When working on an analysis of (presumably) correct input, case marking usually compensates for word order restrictions. However, it is an open question whether that still holds when one has to set out by questioning this basic assumption, as is the case when the input contains constituents in the wrong case. My work will investigate to what extent constraint grammar is a feasible approach in such a setting.