A key factor in the development of Listening Typewriters is the ability to support Very Large Size Dictionaries (VLSD, containing several hundred of thousands words), because any restriction on the vocabulary is a restriction on potential users. This is even more important for inflected languages, such as French. This paper reports some experiments in the development of a speech recognition system using a Very Large Size French dictionary and gives an overview of our decoding strategy and our prototype. The particularity of our system is that it is based on syllable recognition. In the first phase of our research, the speaker is asked to speak with short pauses between syllables. Our next goal will be Continuous Speech Dictation.