<p> PropBank-style (Kingsbury and Palmer, 2002) semantic role labeling has good coverage in several general domains, from the Wall Street Journal Corpus (Palmer et al., 2005) to the medical domain (Albright et al., 2013). The purpose of this project is to explore the efficacy of this labeling schema in the science domain. The My Science Tutor project (Ward et al., 2011) has an abundance of domain-specific data available to evaluate the coverage, portability, and usability of PropBank in sub-domains from the the Full Option Science System (FOSS), such as <i>Energy and Electromagnetism,</i> and <i> Living Systems.</i> The labeler usability will be tested in an off-the-shelf state and compared with manual annotation of the same data, all of which will be taken directly from the My Science Tutor project. A mapping of Propbank- to Phoenix-style annotation will also be devised, and machine learning classifiers for automated output will be created and evaluated. </p>

<p> This dissertation uses a corpus of tokens retrieved from broadcast news stories and print news articles to examine the array of constructions used to encode stative predications in Modern Standard Arabic. A state is defined as a situation that includes its reference time, whether that time is encoding time or another time of orientation. A range of stativity diagnostics are implemented. The constructions analyzed include both those that select for the class of states and those that yield various stative construals of otherwise dynamic predications. The constructions examined range from inflectional constructions to verb-headed phrasal patterns to verbless predicates; a lexicalist implementation of Construction Grammar, Sign Based Construction Grammar, provides a uniform format for representing the constructions as feature-structure descriptions. The constructions include: the p(refix)-stem verb, an inflectional construction exhibiting considerable semantic and syntactic flexibility; participles, including both the Active Participle, which typically yields a progressive reading and sometimes a perfect reading, and the Passive Participle, which yields a perfect reading; non-verbal predicates, which denote various stative relations, including existence, property attribution, possession and deontic modality; and phrasal constructions headed by the auxiliary&nbsp;k?na, which&nbsp;are used to convey past states, irrealis states and resultant states, while serving as a&nbsp;copula in syntactic contexts requiring a copula. A final case study underlines the formal and semantic heterogeneity of the class of Arabic stativizers by examining an emergent idiomatic pattern, the&nbsp;yatimmu&nbsp;construction, which has either a progressive function or a perfect function, depending primarily on subordination. The dissertation shows that in Arabic news narratives, users deploy distinct stative constructions in distinct contexts to convey whatever state is relevant in the context. It demonstrates that constructions convey both tense-based notions (like state ongoing at encoding time) and aspectual notions (state ongoing at the time of another event invoked by the text). In addition, it demonstrates that aspectual constructions are not &lsquo;merely&rsquo; aspectual, but instead have constraints relating to argument structure, valency and subordination.</p>

<p> This study describes the major syntactic features of the Soqotri language, spoken in Socotra Island, Yemen. The Soqotri language belongs to the Modern South Arabian (MSA) languages which with the modern Ethiopian Semitic languages and Central Semitic form the West Semitic sub-branch of the Semitic family (Huehnergard and Rubin, 2011). Based on UNESCO&rsquo;s classification that shows the levels of language endangerment, Soqotri is considered to be a severely endangered language which is spoken only by grandparents and older generations, and is not spoken to young children any more (Moseley, 2012, p. 4). Since Soqotri is regarded as a severely endangered language, it is of utmost importance that it be described and documented.</p><p> The main purpose of undertaking this study is to document the key syntactic features of Soqotri. I present the salient aspects of Soqotri, including basic word order, noun classes, verbs, prepositions, and clausal relations of the Northern Soqotri dialect which is spoken in some villages, such as Qadhub, Hadibo, the Capital City of Socotra, and Hawlef. I also explore the distribution of certain lexical items in Soqotri and their relationship with two linguistic phenomena: negative concord and negative polarity. The study also contributes to typological and historical studies of Semitic languages in the Middle East. </p><p> This study was carried out by interviewing a Soqotri native speaker consultant, who also speaks Arabic, and by examining published data gathered by other scholars in past years. Furthermore, I hope this grammatical sketch will act as a springboard for other researchers who are interested in documenting Soqotri. This thesis also might fill the gap and give insights into the distribution of the negative concord and negative polarity phenomena.</p><p> The results show that Soqotri shares many grammatical characteristics with other Semitic languages, such as verbal and nominal paradigms. Soqotri active clauses basically follow the VSO order. SVO and VOS are also alternatively used for pragmatic motivations. Each type of modifier including adjectives, demonstratives, and relative pronouns agrees in both number and gender with the head noun. According to Giannakidou&rsquo;s (2000) classification of negative concord (NC) languages and based on the data found, we can conclude that Soqotri does not show both types of negative concord. Exclusively, it is a strict NC language. That is, Soqotri NC clauses require the co-occurrence of the sentential negative marker and the <i>neg</i>-word in the clause.</p>

<p> Though the majority of the world&rsquo;s population is bilingual, most of the existing research on child language acquisition has focused on monolinguals. Increasingly, research has begun to investigate language acquisition in bilingual contexts, and has found evidence of both similarity to and difference from patterns found in monolingual language acquisition. One evident source of difference in bilingual language acquisition is interaction, where bilinguals&rsquo; acquisition of one language affects their acquisition of the other language. Interaction has been shown to occur at multiple levels of linguistic structure (syntactic, phonological, phonetic), and manifests in three different patterns: acceleration, deceleration, and transfer. Acceleration and deceleration refer to the rate at which bilinguals acquire some property relative to monolinguals in the same language. Acceleration occurs when bilinguals acquire some property faster or earlier compared to monolingual peers, whereas deceleration occurs when bilinguals acquire some property later or more slowly than monolingual peers. Transfer refers to bilinguals&rsquo; use of a property specific to one language in their other language. While the occurrence of each of these patterns has been demonstrated in bilinguals&rsquo; language acquisition, it is not well understood what causes interaction to occur where and how it does. In this dissertation, I propose that frequency of occurrence and linguistic complexity, both features of the input that are known to affect the course of monolingual acquisition, also direct the appearance of interaction in bilinguals&rsquo; acquisition of language. I present findings from a series of studies demonstrating that differences between languages in frequency of occurrence and complexity of phonological properties influence bilinguals&rsquo; acquisition of aspects of Spanish and English phonotactics in predictable ways. Specifically, greater frequency of occurrence and greater complexity of phonological properties in one language are shown to promote bilinguals&rsquo; acquisition of related phonological properties in their other language.</p>

This dissertation examines imperative constructions within English and across languages. Cross-linguistically, I define the strong imperative to be a unique sentential construction compatible with a direct command and not with an assertion. I show that strong imperatives are not universal: languages exist which can only be characterized as having weak imperatives--sentential constructions ambiguous between assertions and direct commands. The strong imperative lacks both modal elements and elements indicating past tense, and uses formal strategies to mark itself as distinct form non-imperatives. Such formal strategies fall into three types: (i) imperative-marking elements, (ii) the manipulation of subject, and (iii) intonation. Languages use either one of the types or combinations of them to mark the strong imperative. Several implicational universals are drawn from the study, ranging over imperative types, combinations of formal strategies, imperative negatives and the types of subjects. The dissertation proposes to treat the English imperatives as forming a clause type distinct from both tensed clauses and untensed clauses in terms of abstract properties and structures. Two analyses are given, one consistent with Government and Binding Theory (GB), and the other consistent with Extended Categorial Grammar (CG). In GB, imperatives are formally derivable from a single structure underlying both imperatives and non-imperatives only if adjustments to requirements by theta-theory, Case-theory and quantification-variable binding are provided. Negative imperatives are derived by construction-specific rules. In CG, imperatives are taken to be a basic sentence type parallel to declaratives, questions and various other sentence types which all have different clausal structures. The analysis uses lexical types, together with pragmatic issues like the distinction in force between requests and commands, to specify the particular syntactic properties associated with the imperative negatives don't and do not, do and please, accounting for their complex interactions with overt or null subjects. The dissertation also examines the relation between imperatives and tenseless exclamatives--Mad Magazine sentences (MMs). I conclude that MMs and imperatives are not an instance of one sentence type having two distinct pragmatic functions: imperatives have the clause structure of S (TP) and MMs are an instance of S' (CP) structure.

This dissertation formulates, defends, and exemplifies a semantic approach that I call Cognitive Decompositionism. Cognitive Decompositionism is one version of lexical decompositionism, which holds that the meaning of lexical items are decomposable into component parts. Decompositionism comes in different varieties that can be characterized in terms of four binary parameters. First, Natural Decompositionism contrasts with Artful Decompositionism. The former views components as word-like, the latter views components more abstractly. Second, Convenient Decompositionism claims that components are merely convenient fictions, while Real Decompositionism claims that components are psychologically real. Third, Truth-conditional Decompositionism contrasts with various non-truth-conditional theories, in particular with Quantum Semantics. And fourth, Holistic Decompositionism assumes that decompositions are circular, as opposed to Atomistic Decompositionism, which assumes that some primitive basis ultimately underlies semantic components. Cognitive Decompositionism is the conjunction of the following theses: decomposition is Artful (chapter 2), Psychologically Real (chapter 3), Quantum (chapter 4), and Atomistic (chapter 5). As I substantiate these claims, I will be responding to the anti-decompositionist theories of Fodor, Davidson, and Quine.

<p> The Minimalist Program is in principle nothing more than a set of guidelines for cognitive syntactic research. Because of its historical pedigree and foundational assumptions, in practice it functions and is perceived as something closer to a formal theoretical framework. This opens the door to implementational possibilities. Though it is not possible to strictly delineate `"minimalist'" and `"non-minimalist'" linguistic theories by choice of theoretical device alone, it is possible to identify, and make concrete, shared theoretical assumptions and formal devices that minimalist theories draw from. This project surveys the recent minimalist literature and catalogs the most important such devices, unifying them where possible, to build a set of implementational primitives capable of accurately representing a large section of recent proposals in Minimalism. The utility of this approach is demonstrated through the development of a grammar development software toolkit for the Minimalist Program which makes these primitives available to researchers. By implementing theories in this system, it is possible to validate their empirical claims and adjudicate disputes over empirical coverage between competing theories. Sample implementations relevant to the ongoing dispute over the Movement Theory of Control are given. </p>

<p> This dissertation explores dialect diversification in the long-distance New World Plautdietsch speech community. Plautdietsch dialects are traditionally classified as belonging to one of two types: either Chortitza or Molotschna. The traditional dialect classification has recently come under scrutiny because speakers rarely use features exclusive to either type. I propose that variation in vowel production is an alternative way of classifying dialect affiliation. In this project, I analyze both the production of vowels and the production of traditional dialect features used by native Plautdietsch speakers living in North America. This work finds that both the traditional dialect features and the innovations in the vowel system are linked to information about a community's migration history, but the two systems represent different aspects of a community's history.</p>

<p> The current study set out to explore the nature of online lexical access to <i>polysemes</i> and <i>homonyms</i> within L2 learners of English. Two separate experiments were conducted on 30 ESL learners and 30 native speakers of English (as a control group) using the self-paced reading method (SPR) with a view to exploring whether L2 learners of English access the meanings of the lexically-ambiguous words selectively (i.e., only the meaning primed by the preceding contextual information is accessed), exhaustively (i.e., several meanings are accessed concurrently), or in a frequency-ordered way (i.e., the most frequent meaning is accessed prior to the less frequent meanings) during sentence processing. Experiment 1 examined the effect of the lexical ambiguity type on lexical access using three categories of words: 10 polysemes, 10 homonyms, and 10 single-meaning words. Experiment 2 investigated the effect of meaning dominance on lexical processing using 20 polarized ambiguous words with dominant and subordinate meanings to find out whether the frequency of meaning affects the latency of lexical access. The results from the two experiments showed that neither lexical ambiguity type nor meaning dominance significantly affected the processing latencies of the non-native speakers (NNSs) and native speakers (NSs) in context. The results suggest that the nature of lexical access to the meanings of the lexically ambiguous words in L2 learners is selective as long as the word is presented in a sentential context.</p>

<p> Concerning American Parenthetical Expressions in Syntax offers an introductory study of the oddity of parenthetical expressions (or PEs) across American dialects of English from a data-driven, syntactic point of view. CAPES presents the results from over 42,000 speaker judgments of audio files containing spoken utterances with parentheticals. These utterances test the possible interpolation points of four pragmatically defined categories of parentheticals&mdash;Vocatives, Mitigatory PEs, Evidential PEs, and Expletives &ndash; as well as some of the possibilities for multiple PEs appearing in the same utterance. These possible interpolation points have been tested in coordination with complex structures and movement operations. Analysis of these data has shown that there are significant differences in patterns of grammatical interpolation points for each of these categories. Despite the clear distinctions present in these categories&rsquo; interpolation profiles, some positions remain more likely than others to grammatically allow PEs. These positions are, in decreasing order of likelihood, the left edge, the right edge, following the first (i.e. highest) subject, and preceding an embedded CP. The data have also shown sensitivity to movement operations which suggest that they attach at the surface level of syntactic development. Expletives have been proven to stand alone in many respects, being the least likely of all the studied categories to be grammatically allowed in an utterance-internal position. Additionally, though the data show that up to four PEs may be stacked at the left edge, this is only possible when the Expletive is the leftmost PE.</p>