Representations, Derivations, Constraints: The Theory of Vowel Harmony

☞ Many of the changes in generative phonology after The Sound Pattern of English have been driven by considerations of languages different from English. In this class, we study one specific phenomenon, found in many languages but not in (any Standard variety of) English. ☞ Many developments can be seen as an interactions of considerations about representations and considerations about representations ☞ If a language L has vowel harmony, this means that the vowels in L can be subdivided into disjoint subsets S1. . . Sn (called harmonic sets), such that all the vowels within a certain domain (e.g. a word) belong to a harmonic set (quasi-formally :∀word w ∈ LE ∀ vowel v ∈w: ∃ Si ∈ LI , 1 ≤ i ≤ n: v ∈ Si). ☞ For instance: Tangale, a Chadic language spoken in Nigeria, has two subsets of vowels: /i, u, e, o/ and /I, U, E, O, a/, and words such as the following: