This dissertation investigates the morpho-syntactic and semantic aspects of the Korean IHRC (internally-headed relative clause) construction, which has its semantic head within the relative clause, the relative clause being followed by the morpheme kes.As for the status of kes, I argue it is a pronoun, positioned under D or N, depending on whether kes stands alone or is modified by a demonstrative. In either case, the IHRC receives the same semantic interpretation as non-restrictive relative clause.I argue the relative clause contains a full clause including TP, exhibiting a full range of tense and aspect distinction. I provide a morphological analysis of the IHRC predicate, identifying the contribution of each morpheme, and its status with respect to the root predicate.I adopt phase theory to explain the fact that the internal head cannot be embedded within more than one phase boundary. An uninterpretable feature on C in the relative clause is checked against a matching interpretable feature on the internal head--DP or the event of vP--, and the index of the internal head thus percolates up to CP. The modified pro has the same index as the CP and the internal head. An interesting syntactic contrast between kes as D and kes as N is that an island effect emerges in the latter case. This is explained by adopting Johnson (2004)'s account of adjunct islands. When kes is a D, the relative clause is the same workspace as the matrix clause, while kes is an N, the relative clause is the separate workspace.For the semantic aspects of the IHRC, I noted when the semantic head is ambiguous between the subject and another element in the embedded clause, the subject is always favored as a semantic head. This phenomenon can be explained by the minimal link condition: the subject is the closest position to check the uninterpretable feature on C, adapting Lin (2006). Also, split-antecedent readings and VP adjunct head readings are coercion effects of event-interruption reading produced in certain circumstances between the matrix event and the embedded event.

The dissertation explores the morphosyntax of complex predicate constructions (causatives and applicatives) in polysynthetic languages of South Caucasus Georgian, Mingrelian and Svan appealing to the tenets of Distributed Morphology within the Minimalist Program. It shows that the interface between syntax-semantics and morphology of these constructions is not always transparent and mismatches between these components are accounted for by post-syntactic processes, which often result from language-specific constraints on the realization of morphemes per word.

A large domain of linguistic inquiry concerns the nature of words. It is widely thought that words are stored and represented in our minds in a structure termed the lexicon, in which every word has a 'lexical representation'. Researchers conduct experiments and examine intuitions about words to determine the content and structure of the lexicon. One interesting component in lexical representation, for literate speakers, is an orthographic representation for words. It has been traditionally assumed that while this orthographic information is available and useful in such tasks as visual word recognition (i.e. reading) or in writing, orthographic information about words is not necessarily involved in non-visual linguistic tasks, like auditory word perception, or speech production.There has been some research however, which has challenged this notion of the isolation of orthographic information to visual processes. In a seminal study Seidenberg and Tanenhaus (1979) found an influence of orthography in an auditory rhyming judgment task. Subjects were faster to judge as rhyming those pairs which shared an orthographic representation of the rhyme than those who shared only phonology (i.e. pie-tie vs. rye-tie). Additional recent research has confirmed these effects of orthography in auditory perception tasks (Taft & Hambly, 1985; Halle, Chereau, & Sequi, 2000; Ziegler & Ferrand, 1998). Even more surprisingly, some experiments have suggested effects of orthography in speech production (Tanenhaus, Finigan & Seidenberg, 1980; Lupker, 1982; Wheeldon & Monsell, 1992; Damion & Bowers, 2003). These experiments all show facilitated naming latencies for words which share orthographic characteristics with some prime environment. As such, these results can all be explained as effects of orthography on lexical access of words rather than affecting the production process per se.In contrast, the experiments and analyses described in this dissertation show an un-ambiguous effect of orthography on speech production. Orthographic characteristics of word-final sounds, and words themselves are shown to influence the durations of spoken productions of those sounds, and whole words. These effects are robust to the mode of lexical access, whether through experimentally elicited reading aloud of words, or through the spontaneous generation of words in a modified sociolinguistic interview format.

Humanities majors often find themselves in jobs where they either manage programmers or work with them in close collaboration. These interactions often pose difficulties because specialists in literature, history, philosophy, and so on are not usually "code literate." They do not understand what tasks computers are best suited to, or how programmers solve problems. Learning code literacy would be a great benefit to humanities majors, but the traditional computer science curriculum is heavily math oriented, and students outside of science and technology majors are often math averse. Yet they are often interested in language, linguistics, and science fiction. This thesis is a case study to explore whether computational linguistics and artificial intelligence provide a suitable setting for teaching basic code literacy. I researched, designed, and taught a course called Language Processing and the Artificial Mind. Instead of math, it focuses on language processing, artificial intelligence, and the formidable challenges that programmers face when trying to create machines that understand natural language. This thesis is a detailed description of the material, how the material was chosen, and the outcome for student learning. Student performance on exams indicates that students learned code literacy basics and important linguistics issues in natural language processing. An exit survey indicates that students found the course to be valuable, though a minority reacted negatively to the material on programming. Future studies should explore teaching code literacy with less programming and new ways to make coding more interesting to the target audience.

Much research on unaccusativity has been done over the past three-and-a-half decades since the formulation of the Unaccusative Hypothesis (Perlmutter 1978). Researchers have examined the semantics of intransitive verbs as well as their syntax to account for the classification of a verb as either unaccusative or unergative; however, for the most part, a similar conclusion across researchers has been reached that neither the semantics of the verb nor the syntactic structure is sufficient by itself to satisfy certain diagnostics of unaccusativity (Legendre and Sorace 2003). Levin and Rappaport Hovav (1995) argue that the syntactic classification of all verbs is semantically determined; therefore, the unaccusativity or unergativity of a verb is syntactically encoded but semantically determined (Levin and Rappaport Hovav 1989, 1995; Legendre et al. 1990, 1991). However, this does not reveal which semantic properties of intransitive verbs in a given language or across languages determine its syntactic classification. Optimality Theory (Prince and Smolensky 1991, 1993; McCarthy and Prince 1993) is a theory of how constraints interact with one another. The theory does not commit a researcher to using a particular approach to syntactic or phonological structure, but provides a framework for applying constraints and evaluating structural representations (McCarthy 2011). The current dissertation uses an optimality-theoretical approach to split-intransitivity (i.e., unaccusativity) in Swahili and Hittite to demonstrate how variation across languages arises from the distinct constraint ranking that characterizes each language. Additionally, this research suggests the use of a partial constraint ranking (i.e., floating constraint) to account for variable behavior verbs within and across languages. Findings from this dissertation indicate that there are principles that predict the unaccusativity or unergativity for a particular class of intransitive verbs and that there is another class of intransitive verbs whose unaccusativity or unergativity varies across languages. This conclusion supports the moderate form of Perlmutters (1978) Unaccusative Hypothesis.

This thesis presents an analysis of second-dialect acquisition of Pittsburgh English phonological features. Pittsburgh English is the dialect spoken in southwestern Pennsylvania. There are two phonological features unique to the dialect: (i) the [ɔ] realization of the low-back vowel merger and (ii) monophthongal /aw/ (Johnstone et al. 2002). The current study is based on speech data collected from nine participants, native speakers of other dialects of English who now live in southwestern Pennsylvania. This analysis shows that these two phonological features can be acquired. This is the first study to examine Pittsburgh English as a second-dialect. Participants read a word list and a short reading passage adapted from data collection methods developed by Johnstone & Kiesling (2011). I analyzed words containing the low-back vowels and /aw/ using the Praat suite (Boersma & Weenink 2013), an acoustic program that extracts vowel frequencies. These frequencies reveal if speakers produce these vowels as found in their first-dialect or as they are produced in Pittsburgh English, their second-dialect. This analysis revealed that three participants have acquired the merger; of these three, two have also acquired monophthongal /aw/. Furthermore, one participant who lacks the merger has acquired the monophthong. This study also provides an analysis of two speaker variables dialect awareness and gender in the acquisition of phonological features. Participants awareness of the dialect, its features, and any opinions they have about the dialect area were determined through interviews conducted after they provided speech data. I propose that speakers who are aware of the use of monophthongal /aw/ in southwestern Pennsylvania do not produce the feature. I also propose that the presence of the feature correlates with gender, as it is only present in the speech of male participants. However, dialect awareness and gender do not account for the distribution of the merger. These second-dialect findings support previous first-dialect studies of Pittsburgh English (Johnstone & Kiesling 2008; Eberhardt 2009). The analysis put forth in this thesis has implications for dialect studies, as it shows that adults can acquire features of a second-dialect. Furthermore, the same speaker variables that factor into the distribution of first-dialect features are also applicable to second-dialect features. This analysis not only adds to the documentation of Pittsburgh English, but also more generally contributes to the understudied field of second-dialect acquisition.

This thesis presents a comprehensive description and analysis of Gã coordinate systems. Gã is spoken primarily in Accra, the capital of Ghana, and no previous research on Gã has addressed the specific type of locative language described as coordinate systems (Lewis, Simmons, & Fennig, 2013). The current study is based on a corpus of Gã locative descriptions collected through interview-style elicitation sessions with a Gã speaker. Analysis of this corpus has revealed coordinate system morphemes ŋwɛ͂ı͂ up, shĩshĩ down, hı͂ɛ͂ front, sɛ̀ɛ̀ back, nı͂nè-jwurɔ͂ right, and àbɛ̀kú left. In describing the use of these morphemes I use Levinsons (2003) framework of locative language and coordinate systems. I propose that Gã uses the Intrinsic and Relative Frame of Reference types of coordinate systems. Additional characteristics of Gã coordinate systems include the use of intrinsic systems based on an entitys functional characteristics and occasionally on fixed armatures, and the use of left and right in conceptual areas, an extension from the names of hands to relative space in a visual field. My analysis of the Gã data also reveals weaknesses in parts of Levinsons frameworki.e., subtypes of the Relative Frame of Reference cannot be disambiguated, and deictic locative descriptions cannot be considered entirely separate from the Relative Frame of Reference. This analysis contributes to theories of locative language and also contributes to documentation of Gã.

This thesis provides an Optimality Theory (OT) analysis (Prince and Smolensky 1993, McCarthy and Prince 1995) of vowel alternations that occur upon affixation across pronominal prefix-verb root boundaries in Tariana (North Arawak; Colombia/Brazil). Previously, the vowel alternations observed in these environments have been described in terms of linear analysis, as independently-motivated processes under the cover term vowel fusion (Aikhenvald 2003). I propose that the vowel alternations occur as instruments of hiatus resolution. Three alternations, or hiatus resolution strategies, are observable in Tariana pronominal prefix-verb root affixation: no change (or diphthong formation), coalescence, and vowel elision. I propose that sonority sequencing governs the form that hiatus resolution takes. Vowel sequences that rise in sonority undergo no observable change (e.g. /du-éma/ → [duéma] she stands). Because the domain of sonority is syllable-internal (Clements 1990, Rosenthall 1994, Blevins 1995), this analysis carries the implication that sequences exhibiting no featural change between input and output form diphthongs rather than sequences of hiatus. Sequences of falling sonority result in coalescence (e.g. /na-ísa/ → [nésa] they climb) or vowel elision (e.g. /wa-éku/ → [weéku] we run). I maintain that mid-vowel coalescence in Tariana is governed by adherence to ternary vowel height adjacency, as proposed by Gnanadesikan (1997). Furthermore, I submit that coalesced outputs are monomoraic when /i/ is the initial input vowel in the root, because root-initial /i/ carries no mora in Tariana. Vowel elision results in the loss of either the prefix or the root vowel upon pronominal prefixation in Tariana. I propose that the quality of the vowel output, whether it mirrors that of the prefix or root vowel in the input, is determined based on subsistence of the root vowel feature [αback], and an avoidance of the vocalic feature [+high]. The analysis presented in this thesis serves to reinterpret the previous linear analysis of vowel alternations observed across pronominal prefix-verb root boundaries in Tariana. It utilizes OT to functionally unify the vowel alternations, showing that each distinct process occurs in order to avoid hiatus formation.

This thesis presents a comprehensive analysis of the Blackfoot demonstrative system. Previous research on Blackfoot (Uhlenbeck 1938, Taylor 1969, Frantz 2009) identifies sixteen morphemes that make up demonstrative words in the language. I propose a demonstrative template that takes into account the fixed morpheme ordering observed in demonstrative forms. Based on the proposed template, I motivate the analysis of the suffix <i>-ka</i> as encoding motion towards the speaker as this accounts for its position together with the suffixes <i>-ya</i>, <i>-ma</i>, and <i>-hka</i>, each of which encode features of motion or visibility. In describing situational functions of each of the morphemes, I make use of Imais (2003) inventory of spatial deictic features. I present the first analysis of the morphologically analyzable, but heretofore undescribed suffix <i>-o</i> as encoding the geometric configuration feature [interior]. This thesis also offers the first explanation of the syntactic contexts that govern the two identificational suffixes <i>-ayi</i> and <i>-aoka</i>.<br /><br /> Earlier analyses of the Blackfoot demonstrative system focus on the spatial features encoded by situational uses of demonstratives to the exclusion of other pragmatic functions. As a result, the proposals do not address variations in meaning when used in non-situational pragmatic contexts. To address this gap in the literature, I examine non-situational pragmatic functions, as well as symbolic situational demonstrative uses (e.g. deictic projection, wider-context). The result of this study is a comprehensive analysis of the Blackfoot demonstrative system which takes into account both syntactic and pragmatic functions, providing new insights into the meanings of many of the morphemes that comprise the system. It also provides support from Blackfoot for Himmelmanns (1996) claim that there are four universal pragmatic functions of demonstratives, and support for Diessels (1999) claim that situational uses are the basic demonstrative uses from which the others are derived.

Thesis (M.A.) -- Central Connecticut State University, 2008. / Thesis advisor: Carmela Pesca. "... in partial fulfillment of the requirements for the degree of Master of Arts in Modern Languages." Includes bibliographical references (leaves 131-144). Also available via the World Wide Web.