This dissertation aims to examine the following questions: Do all languages have asymmetries in relative syllable prominence? And, how best can a linguist determine relative syllable prominence in a language they do not speak natively, particularly if acoustic correlates of prominence are sparse? To answer these questions, I present results from a series of speech production experiments utilizing the speech cycling paradigm (Cummins & Port 1998), in which subjects repeat sentences in time to a metronome at progressively faster rates. Previous results using the paradigm have shown that speakers from a wide variety of languages align prominent syllables in their respective languages with certain positions within the repetition cycle, namely lower-order fractions of the cycle such as the halfway point. This work has shown that stressed syllables in languages like English, Arabic, and Portuguese behave similarly to “accented” syllables in languages like Korean and Japanese in terms of relative alignment in phrase repetition. But what about languages which display no clear phonetic evidence of stress or accent? The present work addresses this question by examining speech cycling results in the Med0mba, a Grassfields Bantu language with a robust system of lexical tone but which displays little phonetic evidence of stress or accent. Findings indicate that Medumba does display evidence of prominence asymmetries at the word and phrase levels. Specifically, stem-initial syllables are found to exhibit relatively greater prominence than stem-final and non-stem syllables, and syllables occurring in the head position of an intonational phrase are found to exhibit relatively greater prominence than phrase-medial syllables. No differences in the behavior of high tone vs. low tone syllables was found; syllables of either tone could display relatively greater or lesser prominence depending on the prosodic context. Rather, the relationship between tone and prominence is shown to be indirect: as suggested by prior work, high tones can be attracted to positions which are metrically prominent. Crucially, the present work shows that metrical prominence displays similar behavior across languages regardless of whether they have stress. Thus, it is argued, a single notion of metrical prominence (independent of stress) should be applied across languages.