This dissertation focuses on distributional, synchronic evidence for a change that is underway within the clitic clusters of Serbo-Croatian. By comparing observed distributional differences across corpora of the various South Slavic languages, it is shown that Serbo-Croatian is exhibiting a loss of the third singular AUX clitic, something not observed in the more conservative Slovenian, yet a change that appears complete in Macedonian. The corpora used in this thesis were collected off of the World Wide Web using a special tool called a harvester, which was designed by the author. Optimality Theory (Prince and Smolensky 1993) was chosen as the theoretical framework with which to model the data. Because Serbo-Croatian exhibits significant variation in the degree to which the AUX clitic deletes, a unified phonological analysis is argued against in favor of an analysis that appeals to the relevant morphosyntactic features of the clitics with which the AUX is paired. Local conjunction (Smolensky 1993, 1995) is shown to be an appropriate method for expressing the relevant environments. The significant distributional differences between the environments exhibiting deletion drove the choice of the grammar. Further, the distributional differences also show that the change varies across the paradigm and across the population and likely over time. Stochastic OT (Boersma and Hayes 2001) is used to model the variation by expressing the variation using overlapping distributions that exhibit local constraint reranking.

The unique contributions of speaker scholarship to the study of Sahaptian languages in the Columbia Plateau have rarely been considered a domain of inquiry in the field of linguistics. In the present study, I utilize a discourse-centered approach to investigate the ways in which an indigenous language is employed as a resource in the creation of texts. I examine the status of Sahaptin language use in a series of unpublished texts produced by X&dotbelow;ilux&dotbelow;in (Charlie McKay, 1910--1996), a multilingual Sahaptin speaker and scholar from the Umatilla Indian Reservation of northeastern Oregon. I account for the merging of internal indigenous linguistic forms with writing in two occurrences: language documentation and individual expression. The study found that, when a Sahaptin speaker writer transfers his or her internalized language to the written form, Sahaptin discourse and world view play a key role in its outcome.

This study proposes a cognitive network model of the Japanese locative postpositions ni and de (both are translated in English 'in'/'on'/'at') by examining a spoken and written corpus. Specifically, this study examines the implications of Langacker's (1987, 1988, 1999, 2000) dynamic usage-based model by investigating natural discourse data. Within the framework of Cognitive Grammar, it is argued that ni in the spatial domain has two basic usages: (1) stative location and (2) allative (Kabata 2000). The allative ni is a more case marker-like postposition, while de is the prototypical postposition and it marks the conceptualization of a locative setting that references an event (Kumashiro 2000). Previous studies do not discuss, however, the full range of locative usages of de, and do not focus on the subtle semantic difference between these locative postpositions in natural discourse. This study aims to fill in that gap and provide a more complete semantic analysis of the full usage of the locative postpositions based upon natural discourse. This study also addresses functional issues relevant to the actual usage of Japanese locative postpositions in discourse: topic/contrastive marker wa and non-occurrence of postpositions. Overall the results from spoken and written discourse provide supporting evidence for the proposed network model for ni and de: (1) the most frequently appearing senses of ni in the corpus are the prototypical usages of ni, simple stative and allative; (2) de marked different kinds of space from concrete to abstract space, most frequently co-occurring with prototypical dynamic verbs in spoken and written data; (3) some predicates were marked with either ni or de which implies how the speaker construes space in the situation; (4) the main function of wa that follows the locatives in spoken discourse is contrastive, while in written discourse it is a thematic marker; (5) non-occurrence of locative postpositions in systematic and pragmatically as well as functionally predictable. Spoken and written discourse exhibits a broad usage of locative postpositions and thus this study provides a cognitive network model for Japanese locative postpositions.

This dissertation is a model of a unified study of three linguistic aspects: description, theory and application. Tohono O'odham syllable weight is investigated within these linguistic aspects. I propose that O'odham diphthongs are categorized into two groups based on their weight: light (monomoraic) diphthongs and heavy (bimoraic) diphthongs. This is opposed to generally understood diphthong classifications (i.e., falling vs. rising). My conclusion is supported by empirical facts, including morpho-phonological and phonetic phenomena. The generalization is theoretically accounted for within the framework of Optimality Theory (McCarthy and Price 1993, Prince and Smolensky 1993). Moraicity of the light/heavy diphthongs and short/long vowels are analogous, and the larger classification of Tohono O'odham vowels is made: Class L (monomoraic vowels) and Class H (bimoraic vowels). However, the distribution of the vowels depends on syllable type: stressed, unstressed and irregularly stressed. This dependency is accounted for by the following: (i) Predictable moraic structure is not specified in input (this is explained with a proposed constraint, MORAINDISPENSABLILITY or MI); (ii) Moraic specification is in the input only where it must be lexically specified (long vowels and irregularly stressed diphthongs); (iii) Light diphthongs surface as monomoraic due to the loss of a mora. In order to account for the relationship between the unstressed position and light diphthongs, I propose a constraint, POSITIONALDIET, a relative of the Stress-to-Weight Principle or SWP (Prince 1990). Evidence for the diphthong classification comes from an acoustic study of a native speaker and learners of Tohono O'odham. In addition to supporting the classification, the differences between English and Tohono O'odham speakers' treatment of diphthongs is explained with respect to the fact that English diphthongs are always heavy. Finally, implications of this study for Tohono O'odham language teaching are discussed.

This study involved the analysis of a corpus of health brochures in Spanish, contrasting the way this particular genre is formed and understood in the United States and Mexico. The corpus is made up of two subcorpora: a collection of health brochures published in the United States (translated from English into Spanish), and a collection of brochures from Mexico (written originally in Spanish). These subcorpora were analyzed and compared from many different angles, providing a comprehensive view of how the texts are structured and organized. Analysis at the microlinguistic level, grounded in Speech Act Theory, shows differences in the linguistic realization of the persuasive interactional goals in the two text types. At the macrolinguistic level, a factor analysis reveals differences between the two text types in terms of the patterning and co-occurrence of linguistic features serving the overall informational and persuasive functions characteristic of the health brochure. These differences are attributed to the interpersonal relations and cultural and social contexts in which the two text types are embedded. A subset of texts from both subcorpora was field tested with the aim of assessing the impact of the two text types on readers, in addition to readers' perceptions of the texts. Fifty-four Spanish-speaking adults from Mexico participated in this part of the study. Field testing involved a three pronged approach: a pretest-posttest protocol, recall protocol, and interviews. Responses provide a look at the brochures from the perspective of the reader, and shed light on the role of the two text types in the Mexican community. Results of both parts of this study contribute to the fields of translation studies, translator training, contrastive rhetoric, corpus linguistics, genre analysis, cross-cultural communication, and health education and communication.

This dissertation explores the phonetics and phonology of intervocalic consonants in Lheidli, a dialect of Dakelh (Carrier) Athapaskan spoken in the interior of British Columbia. Through a series of studies on Lheidli, I show quantitatively what has previously been noted impressionistically in the Athapaskan literature: intervocalic consonants are remarkably long. The implication of these consonants for the structure of Lheidli is approached from two perspectives. First, I investigate their role from a purely phonetic approach, focusing on their effect on the perceived rhythmic structure of Lheidli. I propose a new model of rhythm, the Enhancement/Inhibition model, in which the perception of rhythm is created by the interplay between primary and secondary correlates of rhythm. Within the proposed model, the Lheidli data show that one of the important secondary correlates is inherent segmental duration, an element that has not yet been considered in the literature. Second, I investigate the role of intervocalic consonants from a phonological approach, focusing on their effect on syllabification. I present the results of a series of studies on the distribution of vowel duration and quality, the distribution of consonant duration, native speaker syllabification intuitions, and the interaction between stress placement and intervocalic consonant duration. Together these studies lead me to analyze Lheidli intervocalic consonants as non-contrastive, moraic geminates. I conclude by discussing the implications of the Lheidli data for phonetic and phonological theory. I argue the duration of intervocalic consonants is encoded in the Lheidli grammar as part of the language-specific phonetics. Furthermore, because this duration interacts with syllabification, it is encoded in the phonology as weight. Although in Lheidli the phonetic duration of intervocalic consonants is encoded in the phonology as well as the grammar, I propose that not all language-specific phonetic properties are specified in the grammar. This is the case for rhythm, for example, which is an effect of other phonetic and phonological factors of the language rather than being a linguistic primitive itself.

This dissertation examines one of the most widely observed grammatical phenomena in a number of languages, the dative subject construction (e.g. John-ni nihongo-ga hanaseru 'John can speak Japanese/it is possible for John to speak Japanese'). The greatest controversy surrounding the Japanese dative subject construction concerns the grammatical status of the nouns (i.e. the ni-marked first NP, John, and the ga-marked second NP, Japanese) in this construction. A number of different linguistic traditions including generative grammar, functionalism and the kokugogaku ('the study of the national language') tradition have examined this phenomenon, and a number of hypotheses have been advanced. However, no comprehensive studies on the phenomenon have yet been done with naturally occurring conversation as the primary source of data. Also very little attention has been paid to the questions of how this construction appeared/developed diachronically. In utilizing pre-modern and modern Japanese discourse data, this dissertation aims to accomplish three goals. The initial objective is to provide evidence that the dative subject construction is, in fact, rarely found in naturally occurring conversation, and the patterns observed in actual discourse are significantly different from those examples found in prior linguistic literature. The second goal is to demonstrate how the occurrences of this construction are similar and/or different depending on discourse types (i.e. spoken language vs. written language; narrative portion vs. conversational portion). The final goal is to offer an alternative to the past approaches. In opposition to the standard account that the Japanese dative subject construction is related to a transitive clause, the NP1-ga NP 2-ga pattern, based on my diachronic and synchronic analysis, I propose that the dative subject construction may have emerged from the existential/locational construction via the metonymic use of ni, which marked a locative-like NP for defocusing its agentivity to avoid the explicit mention of an individual worthy of respect. In view of the results presented in this dissertation, instead of talking of its grammatical status, the so-called "dative subject" ni in modern Japanese discourse may be better characterized in terms of its discourse-pragmatic functions, which derived from its locative nature.

It has long been observed in Mandarin Chinese that when two third tones are adjacent, the preceding third tone is realized as a second tone while the following one remains third tone. This is known as the Third Tone Sandhi Rule (TTSR). The application of TTSR in strings with more than two syllables also takes places, and is subject to the influence of syntax and semantics. The purpose of this study is to examine TTSR applications in both nonsense and meaningful strings. In the former case, this study aims to establish a purely phonological pattern for TTSR application. In the latter, this study incorporates findings in the former and explores how syntactic information is used to establish phonological structure. The analysis is conducted in the Optimality framework, where the mapping between phonology and syntax can be modeled using a set of general structural and alignment constraints. This study supports the proposition that conjunction and disjunction of constraints should be allowed to capture the complexity of syntactic structures. In addition, hierarchical phonological structures are necessary to characterize the influence of syntactic hierarchical structure on phonology. In accounting for TTSR applications, this study proposes that TTSR can be evaluated in different phonological domains but such evaluations need not be conducted in a cyclical manner.

This dissertation explores the phonological systems of Micronesian languages, in search of patterns that are consistent throughout the family and others that are unique to subgroups and individual languages. Using data from Pohnpeian, Mokilese, Pingilapese, Puluwat, Chuukese, Woleaian, Marshallese, and Kosraean, the study focuses on reduplicative morphology as an illustrative window into the phonology of each language. Each language is treated with an Optimality-Theoretic analysis, and consequently the phonology of each is attributed to the same set of principles. However, as a whole, the analyses cannot explain the existence of common and unique patterns within the family. A model of language change called Confluence is presented in order to account for typological trends among related languages. The model acknowledges that learners of languages settle on a grammar based on ambient data; thus, the acquired grammar should approximate the grammar that produces the ambient data. However, a number of perceptual and articulatory factors have the result that the set of data that the learner perceives is unlike the set of data that the ambient grammar would predict. As a consequence, the learner's grammar may differ from that of the ambient grammar. Because of the inclusion of perceptual and articulatory biases in the model, Confluence predicts several outcomes. First, a particular pattern might be resistant to change, or the same innovation might be likely to occur independently in different languages, and either way it is an observable property of otherwise divergent languages. Second, two or more co-occurring grammatical traits of a particular language can be shown to have a common source. Each of the predictions of Confluence is supported in the language analyses. Examples of robust patterns include coda consonant restrictions, moraic feet, and bimoraic reduplicative prefixes. Two examples of recurring innovations are the introduction of final vowel deletion and the avoidance of initial geminate consonants in favor of some other structure. Phonological patterns that co-occur in a principled manner include geminate inventories and morpheme alignment, foot structure and final vowel lenition, and a cluster of properties that consists of length contrasts, word maximality, morpheme alignment, and segment inventories.

This study explores gender differences and accommodation behaviors of Japanese conversation participants in the use of Japanese backchannels. The study utilized three types of dyadic conversations, 5 male single-sex, 5 female single-sex, and 5 mixed-sex conversations, in which the participants were asked to talk about a particular topic for about 30 minutes. Five types of backchannel including short responses such as hai 'uh-huh', reactive expressions such as soo desu ka 'I see", repetitions, collaborative finishes, and resumptive opener, along with nods, were quantitatively scrutinized in terms of their distribution, frequency, and placement. The duration of the listeners' gaze was also calculated in terms of the number of the intonation units on the side of the speaker in question. Analysis of the present study revealed that gender differences in the use of backchannels were more pronounced in the single-sex dyadic conversation. Furthermore, it was found that females in the same-sex dyad group showed distinctive characteristics, compared to other participant groups. For example, the study revealed that females in the same gender-dyad group, overall, showed a relatively frequent use of nods. Accommodation tendency was confirmed in the use of backchannels and the listener's gaze behavior. Overall, women tended to show a greater degree of accommodation in the mixed gender dyadic conversation. Besides, the study pointed out that how men and women accommodate depends on the types of backchannels and listener's gaze. There were cases in which one of the sexes showed convergence and the other showed divergence in a mixed-gender dyadic conversation or vice versa. In the present study, a different accommodation pattern emerged particularly between verbal and nonverbal backchannels. As for verbal backchannels, it was indicated that there were more convergence patterns either from both genders or at least from one of the sexes. With regard to nonverbal backchannels, in this case, nods, however, women constantly showed a convergence pattern, whereas men showed a divergence pattern at all times in the mixed-sex conversation. These results indicated that gender differences in the use of backchannels could be context-sensitive, which is a new finding in terms of the analysis of gender and language.