The phonological representation of vowel height

This dissertation examines three constriction-based models of vocalic height--the Hierarchical Model (Clements 1989, 1991, Clements and Hume 1995), the Bipolar Model (Elorrieta 1996), and the Incremental Constriction Model (Parkinson 1996)--with the goal of determining whether they present better alternatives for the description of vowel height than that which had previously been offered by the Articulator Model (Sagey 1986, Vaux 1994, Halle 1995). After careful examination of the mechanics of these models, the primary conclusion is that the Articulator Model continues to account better for the attested data. Some of the problems discussed with the other models are intrinsic to a constriction-based treatment of vowel height, while others are specific to each model.;For the constriction-based models, the following weaknesses are identified: (1) they are less appropriate for the statement of cross-linguistic generalizations, because of the lack of a direct correlation between features and articulatory movements, (2) they require implicational rules relating aperture features to nodes, such as Pharyngeal, Coronal, and Laryngeal, and, (3) the seemingly simple operations involving the family of aperture features disguise a number of formal complexities. The assumption that total height assimilation and scalar processes should be described by formally simple operations is also questioned.;The principal flaw of the Hierarchical Model (HM) is that it significantly over-generates representations. The number of available representations increases exponentially with the number of vowel heights, precisely when one would expect less representational options.;The Incremental Constriction Model stands in sharp contrast to the HM in that it is a minimal system. In fact, it does not provide enough phonological representations to distinguish all the attested systems. It also lacks the necessary notation to refer directly to some of the natural height classes, which needlessly complicates the statement of rules.;For the Bipolar Model, the main observation is that it introduces an elaborate formal apparatus. It requires operations and constraints that are not used elsewhere in linguistic theory, leading to the conclusion that the advantages to be gained by allowing these devices are not warranted.