Att skriva är silver och tala guld - eller?

Research project
Active research
Project period
2016 - ongoing
Project owner
Department of Swedish, Multilingualism, Language Technology

Marcus och Amalia Wallenbergs minnesfond

Short description

The objective of the project is to find out whether, and if so, how and to what extent, dictation tools (speech recognition) can support the writing process for people with reading and writing difficulties.

Writing by people with reading and writing difficulties

People with reading and writing difficulties often find writing to be their greatest and most demanding challenge. This project therefore focuses on the question of how to provide support to assist tem in overcoming this challenge. The starting point for the project is that, for the majority of those with reading and writing difficulties, spelling is the greatest obstacle to writing. This involves not only making many spelling mistakes, but also worrying about making mistakes and putting great effort into avoiding them, for example, by editing and replacing words. A risk with that is that spelling will take priority over other important aspects of writing such as creating interesting content and a good overall structure, as well as considering the intended genre and type of reader. If this happens, spelling will draw resources from these other writing processes and thereby impact on the development of writing skills in individuals with reading and writing difficulties.

In order to free-up resources for these other processes, it is important to find a way to reduce the demands that spelling places on writing. One way to do this is for people with reading and writing difficulties to dictate their texts with the help of speech recognition. Another is to use a spellchecker. In this project we are focusing the use of speech-recognition tools, but the use of spellcheckers will be investigated as a comparison.

Using speech recognition to facilitate writing

While speech recognition for Swedish a few years ago was both expensive and inaccessible, today systems to write through speaking are built in to most computers, tablets and mobiles. It is therefore of great interest develop our understanding of the effects that dictation has on the writing process, in general, and on the writing of those with reading and writing difficulties, in particular.

At the present time, we lack knowledge about the opportunities, and the limitations, created by using speech-recognition systems for writing, and about the impact they may have on the writing processes and on the finished text when production is by speaking rather than writing. Moreover, it is of interest to know the extent to which the answers to these questions are language dependent.

Different language orthographies are more or less complex, and it is therefore reasonable to suppose that the cognitive resources needed for spelling differs between languages. For example, while Italian is a language with a very transparent and consistent spelling system, the relationship between spelling and pronunciation in English is much less consistent. Swedish comes somewhere between English and Italian in terms of the consistency of the spelling system.

In this project we plan to investigate:

  • how the text is affected by being dictated instead of written (with keyboard) in people with and without dyslexia,
  • the extent to which people with reading and writing can benefit from explicit instruction in writing using speech-recognition,
  • whether English-speaking and Swedish-speaking people with reading and writing difficulties are affected to a similar degree when they dictate their texts.

More theoretically, we are interested in how different language tools, in this case speech and writing, affect text production and how language-dependent this is.

Project partners

  • Åsa Wengelin
  • John Rack
  • Sanna Kraft
  • Fredrik Thurfjell