Motivating Morphological Constituents: A Learning-Based Analysis of Hebrew, Navajo, and Spanish

In both description and theory, a variety of linguistic structures have been posited to account for a multitude of linguistic patterns. In the sub-field of morphology, such units include roots, stems, and morphemes. Previous research that has explored the motivation for linguistic structures and constructs in phonology and syntax has raised questions about the nature of these linguistic structures and constructs. Such work, however, has not focused on morphological constituent structure, a gap which the present dissertation addresses through a focused examination of the evidence motivating morphological constituents in Hebrew, Navajo, and Spanish. The theoretical framework of this dissertation follows a learning-based approach, in which linguistic structure emerges from the application of domain-general learning mechanisms to linguistic data. Under this approach, a constituent is posited when a language exhibits generalizations that are more effectively expressed with it than without it. Therefore, morphological constituents for a given language are motivated through a convergence of linguistic patterns that rely on that constituent for description. In all three languages, the grammatical word and word-internal constituents are motivated. In Hebrew, the word-internal constituents are the root and the stem; in Navajo, they are the stem, the conjunct, and the disjunct; and in Spanish, the word-internal constituent is the stem. A lack of uniformity can be found within each language, such that verbs exhibit more complex constituent structure than other classes of words. In addition, despite the commonality in terminology conventionally employed for word-internal constituents in Hebrew, Navajo, and Spanish, the criteria motivating each of the word-internal constituents are distinct across the three languages, such that, for instance, a stem in one language is not the same entity as a stem in another. A comparison of the findings for the three languages offers implications for linguistic theory. Given that the grammatical word is motivated for all three languages, the present study supports other work in morphology that claims the primacy of the word as a basic morphological constituent. In addition, the distinct criteria that identify each word-internal constituent in the different languages indicate that the word-internal constituents motivated by the learning-based approach cannot be thought of as universal, supporting the theoretical claim that linguistic elements, including morphological categories and domains, should be approached on a language-specific basi