jppj in the latter. (Since any non-syllabic is long as the first member

is deleted after jppj in the latter. (Since any non-syllabic is long as the first member of a cluster, we will have phonetic [p·] in this form regardless of the underlying length of this segment.) But in fact it seems that /kapp-/ 'close' has an alternate underlying form jkap-/, so this analysis is not certain. For example, if we insert the medial /-alakittiye-/, indicating anger or intensity, into h-kap-tone-n-a-1, we may have either h-kapp-alakittiye-tone-n-a-1 or h-kap-alakittiye-tone-n-a-1, the former with pp, the latter with p.5 Given underlying /kap-/, we can derive h-k&p-tone-n-a-1 without postulating syncope after jppj. Indeed, the existence of an alternate (although apparently less common) form h-k8pp-8tone-n-a-l, in which the fa/ of j-aton(e)-/ is stressable and does not undergo syncope, would appear to confirm this approach. 6