Our research investigates phonetic differences between read and spontaneous speech. In particular, we are interested in differences that enable listeners to distinguish between the two speech types. In the experiment described below, we test the hypothesis that the stressed syllable in words with a high information load (the accent-bearing syllable) , contains important information for this perceptual distinction. A listening test, employing a gating paradigm, was carried out, with matching spontaneous and read speech material. The results show that accented syllables indeed contain more information than unaccented syllables. Some acoustic correlates of the accents in read and spontaneous speech are presented.