Vowel reduction in optimality theory

In many languages, the full set of vowels that occur in stressed syllables does not occur in unstressed syllables. This situation commonly results from vowel neutralizations in unstressed positions. For example, the vowels /e,o/ might occur in stressed syllables, but not unstressed syllables, being replaced with the corresponding high or low vowel when unstressed. Such stress-dependent vowel neutralizations are referred to as vowel reduction. This study is based on the examination of several languages with vowel reduction, and attempts to uncover universal cross-linguistic patterns in order to identify the motivations of vowel reduction. These motivations are taken as the basis for a formal analysis of the phenomenon of vowel reduction within the theoretical framework of Optimality Theory.;This survey demonstrates that vowel reduction phenomena neither operate in the same way nor produce the same results cross-linguistically. There are attested forms of vowel reduction which appear to be contradictory in behavior: Some languages will reduce unstressed /e,o/ via raising (to [i,u], respectively), while others utilize lowering of the same vowels (to [a], e.g.), or a combination of raising some vowels and centralizing or lowering others. Yet other attested vowel reduction patterns do not target the mid vowels at all, instead eliminating unstressed low vowels or unstressed high vowels.;The hypothesis advanced here is that vowel reduction is not a unitary process. Instead, there are at least two different types of stress-dependent vowel neutralization, each with its own motivation and formal analysis. Specifically, I posit a contrast-enhancing form of vowel reduction based on avoidance of poorly-contrastive vowel categories in unstressed positions, as well as a prominence-reducing form of vowel reduction based on avoidance of vowel qualities that favor longer articulation times (i.e., lower vowels) in positions that preferentially have short duration. This non-unitary approach to vowel reduction yields several beneficial results. For example, once the two types of vowel reduction are differentiated from one another, it becomes possible to identify certain cross-linguistic patterns and tendencies that occur in association with one or the other (but not both) of these types. Furthermore, differentiation of vowel reduction into two types is useful in predicting directions of vowel neutralizations and contexts where vowel reduction is fully or partially blocked.