From: Bloothooft, G., Dommelen, W. van, Espain, C., Hazan, V., Huckvale, M. and Wigforss, E. [1998]. 'The Landscape of Future Education in Speech Communication Sciences. II. Proposals'. UiL-OTS publications, Utrecht, viii+131 p.

CHAPTER 4

A EUROPEAN MASTERS IN LANGUAGE AND SPEECH

Gerrit Bloothooft
Utrecht University, The Netherlands

Doug Arnold
University of Essex, United Kingdom

Hervé Bourlard
IDIAP Martigny, Switzerland

Martin Cooke
University of Sheffield, United Kingdom

Maria Gligoriadou
Ethniko kai Kapodistriako Panepistimio Athinon, Greece

Steve Isard
University of Edinburgh, United Kingdom

Paul McKevitt
Aalborg Universitet, Danmark

Gabriel Lopez
Universidade Nova de Lisboa, Portugal

Kyriakos Sgarbas
Panepistimio Patron, Greece

Jürgen Trouvain
Universität des Saarlandes, Germany

Briony Williams
University of Edinburgh, United Kingdom

Understanding speech communication requires knowledge of speech as a continuous acoustic signal and knowledge of language as a symbolic structure. These two types of knowledge are characterised by different descriptions and theories, and have traditionally appeared in different curricula such as Phonetics, Speech Technology, Electrical Engineering, General Linguistics, Computational Linguistics and Natural Language Processing (NLP). However, attempts to develop systems that utilise human speech as a means for interfacing with machines have revealed that success requires contributions from both the Speech and the NLP communities. In view of the growth of the Language Industry, there is a rapidly increasing need for people who are able to work in a team that consists of specialists in speech, NLP, and computer science. These team members should be able to work together and should therefore be able to understand the major aspects of each other's specialisms. Hardly any existing academic curricula in Speech and NLP make the connection between the two, even where both are taught in the same institute.

In 1997, a consortium of ten universities [Edinburgh, Sheffield, Essex, Aalborg, Utrecht, Saarbrücken, Lisbon, Patras, Athens, Lausanne/IDIAP] applied successfully under the Socrates programme for the development of a European Masters in Language and Speech. This Masters aims to improve the situation by creating an advanced programme of studies that allows students to obtain the qualifications needed for team work in language industry.

We have created a model for the organisation of the Masters and defined its contents. This paper summarises and explains its major characteristics. One important feature is that participation in the Masters is not limited to the initial project group but open to anyone (both departments and students) who wishes to join. Now that we have set up the basics, a lot of work still has to be done. It would be a great help if we got feedback on the proposals below. They are really new to our field, but may open new ways of cooperation and new ways of improving the quality of education.

2 Model

There is little homogeneity in the educational systems of European countries, as can be seen in an overview of European degrees in phonetics and speech technology (Bloothooft et al., 1995, pp. 182-183). The Masters degree is no exception to this situation. In many countries the Masters as such is not a degree and even in countries where it exists there is a great deal of varietion in its implementation, even among universities. This makes it very hard to think of a European Masters as a legal degree.

At Anglo-Saxon universities the Masters phase normally follows after a Bachelors degree has been obtained. Broadly speaking, there are two types of Masters. The first one is an advanced specialisation in the same area as the Bachelor's degree has been obtained. The second type is the so-called conversion Masters, which should be seen as a broad introduction into some area. The conversion Masters can be entered from a wide variety of studies. In general, a Masters may take one or two years.

Given the maze of implementations and the enormous legal barriers, it may be best to start with our visions and aims. The vision is that we believe that solutions in human language engineering require team workers that have at least a basic knowledge and skills in speech and language. The Masters could be used as a definition of these basics. We do not really care very much when and where this knowledge and these skills are obtained, as long as they are there when a student leaves the university. Interpreted in this way, the Masters would become a quality stamp rather than an official degree. Secondly, we wish to attract more students to the speech and language technology area, notably those who have a background in computer science, electrical engineering, psychology, phonetics or linguistics. This means we have points to offer basics in speech and language and hence a Masters of the conversion type rather than of the specialist type.

If a European Masters is largely defined by contents, the following issues have to be addressed:

• its contents
• the assessment of the courses providing the contents
• the recognition of the European Masters

If a student fulfils all requirements for the European Masters, this will be recognised by means of a certificate awarded by the professional organisations in the field: the European Speech Communication Association [ESCA] and the European Chapter of the Association of Computational Linguistics [EACL]. Both organisations have endorsed the proposal. Since this certificate has no legal status, its value depends on its recognition by prospective employers. Certification by professional organisations is already in use in many other disciplines. The big advantage is that the certificate is independent from the educational systems in the various countries and independent from the legal degree the student will obtain within those systems.

The following types of implementation of the Masters can be envisioned:

• A traditional conversion Masters at one university. This may apply to the Anglo-Saxon universities. In this case the Masters is, for instance, organised as a one-year Master of Science (compare degrees in Cambridge and Edinburgh). The MSc degree and the certificate then coincide, and both will be awarded. In some countries it may be possible to get the Masters organised by a consortium of national universities (all under the same law).
• A collection of courses within an undergraduate programme. This may apply to those countries where a Masters is no official degree, or where a Masters is not presented within a limited period but is a degree related to a three- or four-year curriculum. In several humanities faculties in Germany and The Netherlands, for instance, the contents of the Masters are presented although seldom within a single curriculum. A student can meet the requirements for the European Masters by the appropriate choice of option courses. There is no need for the courses to be clustered within one year, they may be presented at various points during the whole four-year curriculum. The student will obtain the official national degree on the basis of the full curriculum (Magister in Germany, Doctorandus in The Netherlands) and will get the European Masters certificate in addition.
• A collection of courses partly provided by the home university and partly through student exchange with a host university. This model may be regarded as a variant of both aforementioned models in the case of the home university not providing all the courses needed to fulfil the requirements. It is for instance conceivable that the home university only has courses in natural language processing and that the student has to obtain the speech courses elsewhere. Courses provided through the internet may be developed for an international audience and hence reduce the need for actual exchange.

A vital issue of this Masters proposal is the question of when a student has fulfilled the requirements of the Masters. This problem can be solved by identifying the courses that provide part of the Masters contents. A student should follow a collection of courses which together cover the total contents of the Masters. The responsibility of presentation and examination of the courses is entirely with the department of the university that provides them. Although in this way the acknowledgement of courses and the procedure of awarding the student the European Masters Certificate is completely decentralised and not part of the organisation of the European Masters as such, the identification of appropriate courses (accreditation) forms the key to the Masters.

The proposed model requires an accreditation committee. Given the experience some have had with accreditation, this part of the proposal may raise hesitations. However, it should be remembered that the accreditation we need concerns course contents and not full programmes or lecturers. Above all, we do not envision a very detailed accreditation with, for instance, on-site visits. In our opinion course accreditation should imply a judgement on the basis of the following information to be provided by the department:

• a short overview of the contents of the course, the teaching methods used, books or readers used, the number of contact hours, and the number of ECTS [European Credit Transfer System] credits related to the course
• the coverage of the course with respect to the prescribed contents of the European Masters
• exercises, exam questions, project subjects from the last two years

The accreditation agency should consist of four to six experts who are appointed by the boards of ESCA and EACL. The decision of the committee on a course will hold for a maximum of four years. If a course is revised considerably during this period, a new decision has to be requested.

It should be noted that any university can apply for the accreditation of courses (whether or not in the European Union). This brings the European Masters within reach of all universities and all students. A European Masters can never be the privilege of an exclusive group of universities. The proposed model is a flexible and open solution.

If a a department wishes to participate in the European Masters, the following documents have to be provided with the application:

• A description of the coverage of the department's courses with respect to the prescribed contents of the Masters, with the request to accredit them [providing the information that is required for accreditation for every course].
• A proposal as to how a student can obtain missing contents elsewhere through courses that are accredited at other universities.
• A description of the way student exchanges will be realised [in most cases through the Socrates/Erasmus programme to avoid fee problems].
• The procedures that will be followed by the examination board of the department to award the student with the European Masters Certificate.

If there is almost complete coverage within a single curriculum of the European Masters content, one may decide to adapt the programme slightly in such a way that a student automatically fulfils the requirements. It may also be the case that there is almost complete coverage by courses presented in the department or somewhere else in the university (or in a national consortium of universities), but that the courses are not part of a single curriculum but are scattered across several curricula. In this case one should consider increasing the number of option courses within curricula.

If there is only partial coverage within the home university, one should seek universities that can offer the complement (by accredited courses). It is not necessary for these to be foreign universities (but see the requirement for studies abroad below). In all cases the additional courses should also be acknowledged within the home curriculum as option courses. If a student wishes to take part in studies abroad, the exchange could be realised through the standard Socrates/Erasmus programme (for those countries that are part of the European Union or associated with it).

The European Masters consists of a series of taught courses and a period of project work (traineeship). To give emphasis to the European character of the programme, a total of at least three months (whether on courses or project work) should be spent abroad. The European Credit Transfer System has been adopted (ECTS - 60 credit points per year of study). This system is especially useful for the quantification and acknowledgement of the part of study taken abroad. Although the Masters is described by contents only and not necessarily in terms of months spend on a certain topic, the Masters study is thought to require about one full year of studies. From the 60 ECTS credits that cover a year, the total of courses should be worth at least 45 credit points and the traineeship should cover a minimum of 15 credit points (3 months). The terms at least and minimum indicate that longer durations can be scheduled as well.

The traineeship should preferably be teamwork in industry but may be replaced by research work at the home university or at another university abroad. The organisation of the traineeship / research work of a student is the responsibility of the home university of the student. A summary of the thesis resulting from this part of studies must be compulsory presented as a report (equivalent to 4 pages A4) to the international electronic student journal WEB-SLS (run under the responsibility of ESCA, EACL and ELSNET).

The contents of the European Masters have been defined in the following topic areas:

• Theoretical Linguistics
• Phonetics and Phonology
• Human Language Processing
• Natural Language Processing
• Speech Signal Processing
• Statistical Pattern Classification
• Language Engineering Applications

The description of the contents of these areas is given in terms of a summary outline, a series of topics and sub-topics with references to example chapters in textbooks or papers. A full overview can be found at the web pages of the European Masters: http://www.cstr.ed.ac.uk/euromasters

To strengthen the identification of a student with the European Masters, an Easter School is projected as a bridge towards teamwork in industry after the course period. The Easter School will be organised every year in April, possibly in the same place. All students following the European Masters are expected to attend this one-week Easter School, preferably after most of the course requirements have been met. At the Easter School, current research topics will be presented, as well as presentations from industry about the latest developments in LE applications.

Some unofficial examination may take place at the Easter School to get feedback on the level of the students. This examination will not have any consequences for the student's progress.

The Masters will start in the academic year 1999-2000 with the initiating group of universities. A lot of work has to be done to make the proposal work. The progress can be followed at the EuroMasters website (http://www.cstr.ed.ac.uk/euromasters). We would like to encourage everybody who has thoughts on the subject to let us know. These may be supportive opinions but may also concern critical remarks on the concept or the proposed implementation. Departments that endorse the proposals and wish to participate can contact the coordinator of the project for more information (Email ). Since we are still experimentally testing contents, structures, and procedures it may be difficult to join during the academic year 1999-2000. But the European Masters certainly will be open for departmental applications from the academic year 2000-2001 onwards.