• PhD theses
• papers in the field of computer linguistics
• papers in the field of corpus linguistics
• papers in the field of phonetics and speech technology

• Speaker independent robust isolated digit recognition  
  P. P. Boda 
• /r/ deletion in Standard Dutch  
  C. Cucchiarini and H. van den Heuvel 
• Speaker specificity in the occurrence of pitch movements  
  H. Kraayeveld 
• Does lexical stress or metrical stress better predict word boundaries in Dutch?  
  D. van Kuijk 
• Intonation in a spoken language generator  
  E. Marsi 
• Articulation and accentuation: preliminary results  
  T. Rietveld 
• Testing two methods for estimating voice source parameters  
  H. Strik 
• A spoken dialogue system for public transport information  
  H. Strik, A. Russel, H. van den Heuvel, C. Cucchiarini, and L. Boves 

The Department of Language and Speech offers two full-time MA programmes:


1. Language, Speech and Informatics is a four-year undergraduate and graduate programme. Students are trained in the use of computer science in the field of language and speech technology, as well as its academic and civil applications. Courses in this programme cover subjects such as syntax and phonology, but also principles of computer programming, the organization of databases, and basic courses in cognitive science. Apart from these there are more specialized courses, like signal analysis techniques and automatic speech synthesis, which border on the field of physics. In general, the programme aims at training highly-skilled professionals who are qualified for employment in the office automation and information industry, by providing a useful combination of courses in the field of linguistics, phonetics, and computer science.
   
   
2. Speech and Language Pathology is a three-year graduate programme and is organizationally situated in the Department of Language and Speech. A number of faculties participate in this interdisciplinary programme: arts, social sciences and medicine. The curriculum focusses on skills necessary to carry out both fundamental and applied research in the field of pathological and normal language and speech behaviour. The following courses are offered: speech production and perception, psycholinguistics, neuropsychology, anatomy and physiology, instrumentation, syntax, research methodology, speech and language pathology.

The department comprises three sections: Computer Linguistics, Corpus Linguistics and Phonetics/Speech Technology. Thematically, the research topics pursued in these sections are the following:


1. computer linguistics:
   • automatic syntactic and semantic analysis
   • Montague grammar
   • machine translation
   • second language acquisition
   • principle based parsing
   
   
2. corpus linguistics:
   • corpus compilation
   • corpus-related research into lexicology, syntax and discourse
   • CAI-applications
   • tools for the morpho-syntactic analysis of corpora and their subsequent exploitation
   
   
3. phonetics and speech technology:
   • automatic speech recognition
   • text-to-speech conversion
   • voice source modelling
   • prosody
   • speaker recognition and speaker characteristics
   • transcription
   • speech pathology

1. 'Direct government funding': Research activities sponsored by the University of Nijmegen, primarily in the framework of the research programme 'Language and Speech Technology'. Some members of the department also contribute to other faculty wide research programmes called 'Language Theory and Language Description' and 'Language in Social Context'. This funding category covers all permanent staff, plus part of the PhD students (called AIO's) and post docs.
2. 'Indirect government funding': Projects sponsored by the Netherlands Organization
3. 'External funding': Funding provided by the Dutch Ministry of Education or the EU, and research carried out under contracts with commercial companies.

A three-year post doc position was granted to dr. Margriet Jagtman by the university for research in the field of second language acquisition. This post-doctoral project is essentially a continuation of the research done by dr. Margriet Jagtman for her PhD dissertation. Dr. Jagtman will be working at the Department of Applied Linguistics. From the Department of Language and Speech dr. P.A. Coppen and drs. G.J. Hakkenberg wil participate in this research.
  Two PhD theses originating from research completed in previous years were defended in 1995, one on automatic translation and another on linguistic aspects of stuttering. For the time being, there are no plans to continue substantial research in these fields. The PhD project on a natural language interface to a database of chemical experiments was concluded in 1995; the dissertation will be defended in 1996.
  Two PhD projects in the field of speaker characteristics also finished in 1995. The dissertations resulting from these projects will be defended in 1996. Work in this increasingly important field is continuing. Two new projects started in 1995.
  One of these projects is funded by the Foundation for Technical Sciences (STW). In this project two PhD students will work on several aspects of speaker recognition (verification and identification) in two telephone networks (both the wired net and the GSM cellular network).
  Funded under the EC Telematics Application Programme the CAVE project started, which will investigate both fundamental and application-oriented aspects of speaker verification over the telephone. Collaboration with the Forensic Research Lab in this field has been renewed and strengthened. A new collaboration with Dutch PTT has been established.

In 1995 a large investment was made in automatic speech recognition research.
  A three-year post doc position was awarded to dr. Helmer Strik by the Royal Dutch Academy of Sciences (KNAW) for carrying out research into the possibility to exploit articulatory knowledge in automatic speech recognition.
  Our department is one of the key locations in the NWO-funded Priority Programme 'Language and Speech Technology', which aims at combining fundamental research in the fields of automatic speech recognition, natural language processing and dialogue management with the construction of an operational demonstration version of an automated system that provides information about public transport over the telephone. In the framework of the NWO programme, our department will concentrate its efforts on automatic speech recognition.
  Closely connected to the NWO project is the MLAP project MAIS, which also aims at building multi-lingual demonstration versions of telephone-based automatic travel information systems for public transport. Under contract with Philips Dialogue Systems we contributed to the implementation of the first operational demonstration version of such a system for the Dutch language and the Dutch Railway system.
  Funded by the university, a PhD project was started which aims to investigate how prosodic information can be employed in automatic speech recognition.
  Finally, under contract with KPN Research we worked on the recognition of connected digits for Dutch.

Also under contract with KPN Research, a one-year project was started which aims towards the development of software that should make existing software and knowledge for the linguistic phase of text-to-speech conversion accessible to application engineers.

The Polyphone project delivered its final result towards the end of 1995: CD-ROMs were produced with the speech of over 5,000 speakers, representing all regional pronunciation variants in the Netherlands, recorded over the telephone.
  The Onomastica project was concluded as well. Here too, the most important result is a CD-ROM containing phonemic forms of millions of proper names in all 11 official languages of the EU.
  The MECOLB project produced its intended deliverables, i.e., tools for the generation and maintenance of lexica from large text corpora in several EU languages.

Funded by NWO, a one-year project started in 1995, which is carried out in close collaboration with the ENT clinic of the university hospital. The aim of this project is to build a database of simultaneous recordings of oral airflow (measured with a Rothenberg mask) and glottal area (as recorded by means of a flexible endoscope). The data will be used to test theories of phonation and glottal leaks.

In these proceedings we present a general alphabetical list of publications that appeared in 1995, the abstract of a dissertation that was defended in 1995, and abstracts of publications by members of staff. In addition, a number of articles have been included. These articles are reports on recent or current research carried out by members of the department which have not (yet) been considered for publication elsewhere. Because the department is an amalgamation of three sections, it has been decided to present the abstracts of publications accordingly. They can be found under the headers Computer Linguistics, Corpus Linguistics, and Phonetics/Speech Technology.