Sevillian Spanish is undergoing a metathesis change in /s/-voiceless stop sequences, whereby coda /s/ debuccalizes to [h] and then metathesizes with the following stop (e.g. /pasta/: [pahta] –> [patha]).  While the phonetic reality of this change is well-established (e.g. Ruch & Peters 2016), the phonological behavior of the resulting [Ch] sequences has not been investigated.  In most languages, [Ch] sequences are aspirated stops.  It has been proposed that Sevillian [Ch] sequences may be coalescing into single segments (O’Neill 2009), and this would represent an oddity in the realm of sound change.  In this talk, I present results from a series of experiments testing the underlying representation and possible causes of this metathesis change.  Behavioral evidence from two perception tasks suggest that Sevillian listeners still treat [Ch] sequences as underlying /s/-voiceless stop sequences: they map [h] in [Ch] sequences to an /s/ on a preceding word, and they treat syllables preceding [Ch] as if they were still closed by [h] for the purposes of stress assignment.  Finally, results from a cross-linguistic ABX task show that the cause of laryngeal metathesis in Sevillian (or in other languages) is not likely to be perceptual.  The results have implications for our understanding of segments, clusters, laryngeal metathesis, and the relative rarity of preaspirated segments cross-linguistically.