Specifying Accent Marks in French Text for Teletext and Speech Synthesis

Abstract A system to insert accent marks in unaccented French text is described. It has applications to teletext, speech synthesis, and orthographic pedagogy. The placement of French accents is governed both by syntax and by orthographic patterns of letter sequences. This paper concentrates on exploiting regularities in letter contexts to specify accent placement. The structure of the rule system is given, as well as some examples of how the system operates. Testing used a 50,000 word dictionary and a 31,750 word text. A comparison of this method with a simple look-up approach is made, from the point of view of accuracy and computer memory and computation.