ABSTRACT
Clements (1985), citing work by Mohanan (1983), notes that in current phonological theory there should be 'three common types of assimilation processes in the world's languages': total assimilation, expressed as spreading of the Root node; partial assimilation in which a class node spreads; and single-feature assimilation in which a single feature spreads. This chapter tests these predictions against the laryngeal features, particularly for the feature [constricted glottis] ([c.g.]), which characterizes ejectives. Total assimilation of ejectives is confumed in §2.8; class node assimilation, where the class node is the Laryngeal node, is confumed in §§2.4 and 2.5, and for the Place node in §2.7; and single-feature assimilation is confumed for [voice] in §2.3. However, not all features appear to behave independently, since ejectives behave in a very asymmetric way from how the theory predicts. I have found no evidence to show that ejectives spread [c.g.] phonologically, while other laryngeal features remain inert (thereby proving that only the single feature [c.g.] spreads and not the whole Laryngeal node). This raises the question whether the features are adequately characterized and organized, or whether this should be reconsidered.
