next up previous contents index
Next: Language models Up: Spoken language lexica Previous: Recommendations on lexicon construction


In this chapter, an overview of applications, requirements, lexical information, lexical representation and lexicon structure for spoken language systems  is given, with particular reference to the differences between lexica designed for spoken language as opposed to written language processing. Specific recommendations were provided at the relevant points.

The state of the art for spoken language lexica is such that at the level of word forms, many basic standard techniques have been developed and are widely used. Large-scale spoken language lexical resources are required for current research and development with both statistical and knowledge-based technologies, and need to function as. reference sources of standard, stylistic and regional pronunciations, and in general vocabularies which are characteristic of spoken language, including, for instance, discourse particles,

Spoken language lexical resources in the form of actual lexical databases and tools for constructing them are, however, sadly lacking - even more so than for written language, partly because of the specific complexities of spoken language, and partly also to the fact that the construction of spoken language lexical databases is highly labour-intensive, therefore expensive, and consequently proprietary and in-lab developments are not often made generally available.

The major requirement in this area is simply that this gap urgently needs to be filled. Local and national project work can go some way to meeting the need. In view of the labour-intensive character of lexicon construction and, the feasibility of constructing lexical representation languages and lexical acquisition tools which can be deployed in multifunctional and multilingual environments, cooperative international initiatives are essential.

An important area which has not been touched on is the question of how spoken language lexica relate directly to the human user. This question may be answered in terms of two main areas:

  1. Cognition and the spoken language lexicon. 
  2. Ergonomics and the spoken language lexicon. 

The first answer is a theoretical perspective, pertaining to the theoretical sense of the mental lexicon, lexical access and lexicon acquisition strategies of the user, which the lexicon of a spoken language system , in some sense, models [Marslen-Wilson (1989)].

The second answer touches on a more practical area, namely how the role of the lexicon in a spoken language system  can be defined in such a way as to permit the most ergonomic, user-friendly development and deployment of complex spoken language systems . The former concept is an essential source of ideas and criteria for the latter.

Treatment of these areas, taking extensive work in experimental psychology and psycholinguistics  as well as user requirements into account, would be premature at present, but may be predicted to play a prominent role in future research and development.


next up previous contents index
Next: Language models Up: Spoken language lexica Previous: Recommendations on lexicon construction

EAGLES SWLG SoftEdition, May 1997. Get the book...