We used an annotated conversational French speech corpus to 1. investigate whether the intonational rises that occur at the beginning of French content words in read speech (APRs) are also present in spontaneous speech and therefore available as cues to word segmentation and lexical access, and 2. test two measures of characterizing intonation patterns using automatically extracted F0 and time values. The two measures tested both proved problematic: they were sensitive to the segmental composition of the critical region. We found no evidence that APRs are reliably present in the corpus as a whole, although we suggest that they may be present in particular types of conversational speech.