This study is aimed at a better understanding of the perception of syllables. As the traditional view seems to associate syllable perception with segmental cues that result from local (i.e. present only within or adjacent to the syllable) supralaryngeal events, we are particularly interested in whether non-segmental and non-local laryngeal information contribute to syllable perception as well. Existing works on Indo-European languages show that local stress patterns and global (i.e. non-local) speech rates provide perceptual cues to words and phonemes. While we believe that the effects of the global speech rate hold across languages, based on the long-developed notion of language-specific perception, we expect that lexical tones, rather than stress patterns, serve as an important local non-segmental cue in tonal languages. We conducted a perception study on Mandarin to investigate whether tonal f0 patterns and speech rates interfere with spectral information in determining the number of syllables in an utterance. F0 contours were generated using the qTA model (Prom-on, Xu & Thipakorn, 2009). Our results show that the perceptual number of syllables depends on the perception of tonal f0 patterns and speech rates to a substantial extent. Combining our findings with prior claims (Olsberg, Xu & Green, 2007), it appears that a variety of cues - segments, lexical tones, and speech rate - compete in perceiving Mandarin syllables. In relating this study to the existing works on word segmentation, lexical access, and phoneme identification,we find that the language comprehension system integrates local with global, supralaryngeal with laryngeal information, in perceiving linguistic units - not only words and phonemes, but also syllables.