This dissertation is concerned with the morphology and tonology of Shona, a Bantu language spoken in Zimbabwe. These two topics are treated together because, as I argue, they are inextricably linked: the tone rules are sensitive in various ways to morphological structure, and this sensitivity in turn provides crucial information about that structure. After a brief introduction to the language, we turn to questions about the structure of words in Shona. I argue that Shona word structure is ordered into two levels, stem and word, which can be characterized by a simple context-free grammar. But the word so defined is not the traditional word that has been assumed in the orthography and in most linguistic studies of the language. I argue that the grammar is complicated and crucial generalizations are lost if this traditional word is considered to be a syntactic or morphological unit. Rather, it is to be analyzed as a phonological word, i.e. a derived domain for phonological rules. I present an algorithm for parsing Shona phrase markers into phonological words, within the universal theory of phonological domains developed in recent work by E. O. Selkirk. The domains defined in the second chapter--stem, word, phonological word and phrase--are then exploited in an analysis of Shona tone patterns in Chapter Three. Tone patterns from various dialects are examined and compared: Zezuru, Karanga, Manyika, and Ndau. It is shown that insight can be gained into these patterns if we assume the domains defined in Chapter 2, together with some general constraints on phonological representations including in particular the Obligatory Contour Principle. Given these assumptions, it is possible to reduce the basic tonology of Shona to a single rule of assimilation and two rules of dissimilation.

This work is intended as an investigation into elliptical phenomena in natural language. It is argued that at least two major classes of elliptical constructions must be distinguished, on the basis of the presence or absence of their major phrasal heads. The recovery of the missing material in the class in which the relevant heads are missing (H('(TURN)) class) is of a syntactic nature, and this along with other 'characteristic properties' of H('(TURN)) constructions follow as a direct consequence of the omission of their syntactic heads. In constrast, constructions in the H+ class are pronominal in nature, and their characteristic properties follow from the fact that pronominals may be interpreted either in the syntax (e.g., as bound variables) or in the discourse representation. An account of these constructions is proposed within the Government-Binding framework, and consists of four main components: (i) a 'defective' X-bar schema, which allows for the base-generation of H('(TURN)) constructions, (ii) a licensing principle on D-structure representations, which constrains the output of the defective rule schema, and (iii) a process of elliptical reconstruction at LF, which applies to both H('(TURN)) and H+ constructions, and (iv) the reintroduction of the more general notion of 'recovery of content' to subsume the more narrowly defined notion of 'syntactic identification.' It is argued that this proposal has several desirable consequences. In the first place, it provides a strictly syntactic basis for the H+/H('(TURN)) classification, and derives the distributional and interpretive properties of these constructions from the interaction of this syntactic fact with existing principles in the theory. It accounts for interesting similarities between null arguments (pro) and H+ phenomena. It sheds light on various aspects of the interpretation of both elliptical and overt pronominal elements. And finally, it makes predictions about the range of variation in the manifestations of H('(TURN)) and H+ constructions that one may expect to find in natural language. The linguistic data in this work is drawn primarily from English, but constructions from French, Brazilian Portuguese and Chinese are considered as well.

This thesis is essentially a comparison of various optional infinitive clauses in English. The purpose clause, shown in (1), and the rationale clause, in (2), have literatures of their own; the clause in (3) has been discussed under the name "objective clause" by Faraci (1974), though later authors have identified it with the purpose clause. (1) Sue built the extra room (to hold her sewing supplies); (2) Mary sold her car ((in order) to pay the rent); (3) We brought Sam along (to amuse the children). I argue that there is indeed a distinction between a purpose and an objective (which I call a "goal") clause; in fact, at the VP level we may distinguish at least five different optional infinitive constructions. (4a-e) show infinitives which I refer to, respectively, as purpose, result, goal, exchange, and stimulus clauses. (4) (a) Sue built the extra room (to hold her sewing supplies); (b) John awoke (to find the fire had gone out); (c) Sam came along (to look after the children); (d) They gave Sue ten dollars (to pose with a cobra); (e) Mary blushed (to recall Tom's importunities). Rationale clauses, of course, are S-level infinitives and do not belong to the paradigm shown in (4); still, there is more than one kind of infinitive possible at the S level. (5) compares a rationale clause ((a)) with an "outcome" clause ((b)): (5) (a) I gave Scruffy a biscuit ((in order) to keep him quiet); (b) Mary escaped (only to be recaptured). Each of these seven infinitive clauses has a different semantic relationship to the main clause; however, I claim that none of them has inherent semantics or even a unique syntactic structure. On the contrary, I suggest that all adjoined infinitives are basically the same infinitive, with its semantic force determined by its syntactic context: the attachment point of the adjunct, the structure of the main clause, and the antecedent ("controller") chosen for the adjunct's subject position.

A comprehensive, coordinated analysis of Finnish syntax is attempted in this dissertation. The following areas receive special attention: word order, grammatical case marking, null subjects, binding, and non-finite constructions. The basic framework used is the Government-Binding Theory, but a number of revisions are introduced in order to account for the Finnish data. The theory of levels of representation is enriched by allowing for postcyclic operations in syntax. A theory of structural default case is developed which includes a level of postcyclic genitive case percolation. The current theory of movement is modified so as to allow for a postcyclic movement operation (following WH-movement and Topicalization); this modification also involves revising Trace Theory. It is argued that Finnish has fixed syntactic positions which need to be licensed at specific points during the derivation. This requirement has the effect of producing instances of obligatory movement in order to fill a syntactic position, giving rise to word order variation. A strong form of the notion of Structure Preservation is supported by the data. The theory of default case that is developed in this dissertation provides a way of studying the interaction between morphological case assignment, subcategorization, syntactic positions, and movement. Developing this theory involves reducing the theory of Abstract Case into aspects of the other components of the grammar.

The goal of this dissertation is to examine the nature of Japanese phrase structure and the related parametric differences in syntax between English and Japanese. In chapter 1, I will present some basic concepts and principles of the framework of the principles and parameters theory within which the present study is conducted. In chapter 2, I will argue that Japanese clauses (S and S') are IPs and CPs, respectively, and that the "subject" of S is in the specifier position of IP, and is assigned nominative Case by Infl. I will then show that some of the observed syntactic differences between English and Japanese follow from the difference in the value of the directionality parameter of head-government and its interaction with other principles of UG. Chapter 3 is devoted to the examination of the internal structure of "noun phases" in Japanese. I will argue that noun phrases in Japanese are DPs, i.e., maximal projections of the functional category D. In chapter 4, I will propose an analysis of the so-called "major subjects" in Japanese. I will argue that major subjects originate in the specifier position of the "subject" DP, and raise to the specifier position of the higher empty Infl. I will show that the properties of major subjects follow from other independent properties of grammar in this analysis.

This thesis investigates issues in the phonology and morphology of the Navajo verb. Navajo, as an Athapaskan language, has been considered a canonical example of a slot-and-filler template, a typologically odd morphology. An alternative treatment of the verbal structure is offered: the verb is shown to have a bipartite structure consisting of two constituents that coincide with the universal categories 'Verb' and 'Infl'. This analysis has the benefit both of bringing Navajo into discourse concerned with the cross-linguistic investigation of principles governing word formation and of providing evidence for a configurational theory of morphology. The phonology provides the bulk of the arguments for the bipartite structure. Prosodic phenomena provide a consilience of arguments for the structure; syllable theory predicts the existence of an internal word boundary in the verbal template, segments at the boundaries shows the edge effects of extraprosodicity. An elegant account of a pervasive but illusive epenthesis process can be given based on this structure. Several traditional problems involving the characterization of the phonology required to account for the morpheme concatenation of the template are solved. The bipartite structure is then exploited in an examination of the segmental phonology. The alternations found on the initial consonants of the verb stem are taken up. An association of glides to a pattern of fricative phonology is uncovered which allows the unraveling of intricate glide/consonant and glide/vowel alternations on verb stems. It is proposed that two fricative types exist in the language, type A, the regular fricative, and type B, a fricative type that shows glides as reflexes. This analysis, extended to glide alternations in the Infl constituent, shows promising results towards unraveling the phonology of the verbal prefixes.

A Noun Phrase must have some kind of thematic role in order to be interpreted in a sentence. These thematic roles are considered to be assigned by a predicate with an argument structure in the Government and Binding framework, which is assumed in this dissertation. An argument structure specifies what thematic roles a predicate has to assign. This dissertation is concerned with how such thematic relationships can be related to structural relationships. It is argued by Baker (1988) that identical thematic relationships between items must be represented by identical structural relationships. This direct correlation between thematic relationships and structural relationships is a simple and ideal tool for a child acquiring a language. Whenever two items which are thematically related appear, their structural relationship is the same anywhere. Baker's claim argues for the Unaccusative Hypothesis of Relational Grammar, which claims that the subjects of some non-agentive verbs originate in the object position. The claim also argues against lexical derivation of complex predicates. This is because the thematic relationship in a clause must be maintained even when the clause is embedded as a complement clause. As a consequence, Baker's claim argues that the affixation of complex predicates takes place in syntax. We will argue that there is a syntactic operation called Incorporation in Japanese, which observes the structural relationships between theta-related items. The types of incorporation in Japanese that we discuss are Noun Incorporation and Verb Incorporation. We further argue, considering the interaction among complex predicates, that Incorporation can take place at different levels. Some cases of Incorporation take place in the mapping from S-structure to D-structure, and other instances of incorporation take place at the level of Phonological Form (PF).

A system of three laryngeal features, called (voice), (aspiration), and (glottalization), is argued to be the correct representation of laryngeal distinctions in consonants, based on evidence from possible consonant systems and phonological rules. These features are shown to be privative or single-valued. It is shown that the process of voicing assimilation must be analyzed without use of a feature (-voice), as a combination of neutralization and spread. Neutralization and spreading involving aspiration and glottalization are also analyzed. Neutralization is argued to be a result of a positive well-formedness constraint, the Laryngeal Constraint: a Laryngeal node is only licensed in the configuration in (i).(DIAGRAM, TABLE OR GRAPHIC OMITTED...PLEASE SEE DAI)In any other position the violation of well-formedness will be repaired by delinking the node. In many languages word-internal syllable-final consonants neutralize, but word-final ones do not. This is accounted for by Final Exceptionality: like Final Extrametricality, which allows otherwise forbidden segments at word-edge, otherwise forbidden feature specifications are allowed at word edge. This makes the correct cross-linguistic prediction: word-final neutralization always correlates with word-internal neutralization, but word-internal neutralization can occur without word-final neutralization. The Constraint can be further restricted: languages may neutralize laryngeal distinctions in obstruents and not in sonorants, or may neutralize only one laryngeal feature rather than the entire Laryngeal node. It is shown that the latter is not counterevidence to the existence of the node, since the majority of cases neutralize all distinctions. Sonorants are argued to be unspecified for (voice) until late in the phonology. They do not participate in neutralization of (voice) or in voicing spread until the postlexical phonology; voiced sonorants are the output of neutralization, and so must be the laryngeally unmarked type. Voiceless sonorants are analyzed as aspirated, which is necessary if there is no (-voice) feature, and has supporting evidence from phonological rules.

This work examines a range of phenomena in the phonology of Modern Irish, in particular the initial consonant mutations, distinctive palatalization and epenthesis and compensatory lengthening. A convergence of evidence from these different domains argues for the theoretical frameworks adopted, in particular those of Feature Geometry Theory and Moraic Theory. The articulated structures of Feature Geometry Theory allow for a more principled account of the initial consonant mutations than is possible within other theories. The epenthesis and lengthening processes may be accounted for only within a prosodic theory based on syllable weight. Issues bearing on Underspecification Theory are also addressed. In Chapter 2, the initial consonant mutations that express morphological distinctions in Irish are discussed. The following claims are argued for: First, these mutations are the result of rules, rather than morphological affixation. Second, these rules are blocked by linked heteromorphemic structure, however, this Inalterability effect cannot be captured by current proposals in the literature. Rather, a more general condition on structural conditions on rule application is proposed. Third, vowel stems in Irish are argued to have underspecified onsets. The range of prothetic consonants that occur preceding vowel-initial stems in certain morphological environments is shown to be the surface realizations of these onsets. Chapter 3 focusses on the Place features and palatalization. It is argued, on the basis of evidence involving the initial consonant mutations and assimilation, that the primary and secondary place features must be represented independently. Of further interest are data involving consonant-vowel interaction which argue for context-sensitive Radical Underspecification, whereby lexical and default feature values are relativized to the categories for which they are defined. In Chapter 4 a unified prosodic account is given of a number of epenthesis and vowel-lengthening processes in Irish. The epenthesis process discussed is not conditioned in an obvious syllabic way, nor is the vowel-lengthening a typical case of compensatory lengthening following deletion of segmental material. Rather, both processes are the result of the transfer, and thereby the preservation, of prosodic weight within the syllable.

This dissertation examines the role of the stricture features (continuant), (consonantal) and (sonorant) in phonological phenomena involving place of articulation. It is argued that these features play a central role in such phenomena, a fact that requires an explanation in feature geometrical terms. Arguments from topics as diverse as place assimilation, stricture in complex segments, place-stricture interaction in cooccurrence restrictions, spirantization, and the phonology-phonetics mapping converge in support of a new geometry grouping (continuant) and (consonantal) with the articulators, a configuration referred to as the 'articulator group'. Chapter 1 introduces the background issues. Phonetic considerations are also explored, and found to lend support to the proposal. Chapter 2 examines an asymmetry in the triggers involved in nasal place assimilation: In languages where there is assimilation to continuants, there is assimilation to stops, while the reverse entailment does not hold. This asymmetry is explained assuming the articulator group and a theory of marking conditions and Structure Preservation. Chapter 3 turns to the representation of complex segments, focusing in particular on Kabardian, arguing that the articulator group is required in order to adequately represent such segments. In chapter 4 the role of (cont) in place assimilation by liquids and obstruents is explored. The few clear cases of such place assimilation do not challenge the proposal; some directly support it. Some instances of Coronal linking are more troublesome, suggesting the possibility of more structure within the articulators. Chapter 5 takes up some important further issues, including the implications of the articulator group for a theory of Coronal underspecification and for the representation of laryngeal segments. The examination of these issues provides further support for the proposal. Chapter 6 treats the role of stricture features in consonant cooccurrence restrictions. A revised form of the Obligatory Contour Principle is advocated, and further implications for the articulator group are examined. Chapter 6 concludes with a discussion of the use of class nodes in Feature Geometry.