Pre-Fortis Clipping is a phenomenon in which vowels are shorter before voiced codas than before voiceless codas. This study tests the hypothesis that the voicing feature in the coda of a syllable can also affect the duration of the initial sound in that syllable, in a manner similar to Pre-Fortis Clipping. Following work done by Hawkins and Nguyen (2003), four measurement studies concentrate on the sonorant consonants /l/ and /n/. Results of the study indicate that American English has this phenomenon as a possible effect for some speakers, contrasting with previous studies of British speech.

In this study, I present analyses of various aspects of Karitiana, a Tupi-Arikem language spoken in the state of Rondonia, in the Amazon region of Brazil. These analyses range in both methodology and scope, but are unified by the goal of elucidating articulatory, perceptual, and morphosyntactic patterns in the language. These patterns crystallize during the course of each of two parts of the dissertation. In the first part, I focus on the sound system of Karitiana. The investigation of the sound system includes detailed analyses of the phonemic inventory of the language, as well as basic phonological processes. However, the investigation also includes acoustic, quantitatively-oriented examinations of the vowel system, stop-vowel sequences, and patterns of velar lowering, among other phenomena. These studies focus on basic articulatory gestures in the language. Basic perceptual patterns, related to the typologically-unusual patterns of velar lowering in the language, are also considered in Part I, via the discussion of a speech-perception experiment carried out among the Karitiana. In Part II morphological and syntactic patterns in Karitiana are examined, within a functional-typological framework. Part II includes a grammatical sketch of Karitiana morphology and syntax, as well as more detailed studies of two basic aspects of Karitiana morphosyntax, namely grammatical relations and voice phenomena. I demonstrate that in some cases the latter sorts of phenomena reflect basic conceptual patterns, associated with the construal of event types, which are evident in the grammar of Karitiana.

A reference grammar of the endangered indigenous Eastern Khanty dialects of Vasyugan and Alexandrovo of the Uralic language family has been developed. The study bases on the corpus of natural narrative discourse, and is set in a general cognitive-functional, usage-based model of language. The description addresses the main patterns of the Eastern Khanty language system and offers some typological contextualization of the reviewed language data. The description covers the issues in phonology, word-classes, morphology, syntax and semantics of simple and complex clauses. In the area of phonology, such systematic features as robust backness vowel harmony and consonant-vowel harmony are analyzed in the articulatory gesture framework. Morphologically, the system is agglutinating with suffixation dominant in derivation and inflection. Syntactically, Eastern Khanty patterns as a typical SOV language. Occasional non-prototypical features include non-canonical argument marking along ergative pattern against the general background of Nom-Acc system of GR organization. In mapping of the pragmatic functions---to semantic roles---to grammatical relations, Eastern Khanty shows strong preference towards Topic-initiality, typically mapped onto Agent semantic role. This preference remains dominant in detransitivisation operations, where the prototypical mapping is altered towards Topic-Target-S that generally has to do with the parenthetical demotion of pragmatic status of the Agent referent and promotion of the non-Agent. Analysis of Eastern Khanty complex clauses reveals robust use of finite and non-finite (participial, infinitival and converbial) constructions as relative, adverbial and complement clauses in typologically common strategies of clause-linking. Traditional discrete differentiation into subordinate and coordinate types based on morphosyntactic criteria appears inadequate, divorced from the structural diversity of the observed complex clauses. Cognitive-functional approach is used instead, implying a universal way of construal of linked events, appealing to cognitive relations between states-of-affairs, rather than particular structural means. This results in a continuum of combinable features locating each clause in relation to either the subordinate or coordinate prototypes. Clause-linkage strategies are related to the pragmatic differentiation of information in utterances, with pragmatics, information structuring aspects being at the core of the distinction between the subordination and coordination.

This study offers a new approach to grammatical constructions that express futurity in Danish, Dutch, English, German, and Swedish. Future constructions develop out of lexical elements whose meanings persist to some degree in modern usage. Future constructions thus convey not only future time reference, but also modal meanings of volition, obligation, or possibility. This multifunctionality has been a challenge for previous accounts that aimed to delimit their function to either tense or modality. The present study aims to overcome this debate and views future constructions as meaningful units of language, not as mere paradigmatic alternatives to temporal or modal categories. The present study develops a methodology that provides the study of meaning with a strong empirical basis in the form of large computerized text collections. In order to characterize the meaning of a given future construction, the present analysis investigates the types of main verbs that typically co-occur with it in actual usage: If a given construction typically occurs with intentional verbs such as write or speak, its meaning will differ from constructions that typically occur with verbs such as rain or increase. The same methodology is applied to the historical study of future constructions: If a given construction tends to co-occur with different main verbs at subsequent stages in time, this is indicative of a semantic change. The results of both the synchronic and the historical investigation challenge results of earlier studies. For instance, the English future construction with be going to has been commonly assumed to be a translational equivalent of Dutch gaan. Evidence from modern usage data shows that this is not the case, since both constructions are used to refer to different types of future events. With respect to the historical development of future constructions, several developmental paths have been proposed for constructions deriving from verbs of motion, obligation, and other lexical sources. For example, verbs of motion are supposed to become markers of intention before they acquire temporal meaning. This study presents historical data from a Swedish future construction with the verb komma 'come', which has developed in an entirely different way. Overall, the comparison of future constructions across languages and across different periods of time is intended to develop an understanding of these constructions as meaningful units. While many aspects of these constructions are idiosyncratic, and thus account for the controversies that have surrounded them, some aspects are invariant across the investigated languages, and thus seem to be typical of future constructions in general. The study also yields new insights into the workings of grammatical change, especially regarding the function of co-occurring lexical material in the historical development of constructions. Methodologically, the present study breaks new ground as it empirically tests general tenets and specific proposals regarding grammatical change on the basis of primary usage data.

This thesis investigates the knowledge which Chinese speakers must have that enables them to produce and comprehend Chinese sentences because grammatically Chinese provides little formal marking for syntactic functions such as subject, object, etc. and participant roles such as agent, patient, etc. The present work also presents a model for the representation of such knowledge. Using a cognitive approach which stresses the knowledge of users and the conceptual structures of the linguistic system, this study argues that Chinese speakers must know the conceptual relationships between processes and participants when processing Chinese sentences. Three types of knowledge are posited for the understanding of these relationships: knowledge about the world, linguistic knowledge, and pragmatic knowledge. A classification of Chinese conceptual processes is done according to the conceptual criterion PERIODICITY. Four basic types of conceptual processes are derived: State, Status, Action and Event, each of which corresponds to some syntactic properties and a different type of conceptual relationship. The categories of processes and participants have their hierarchical structures which are composed of two types of relationships: subordination and part-whole. Knowledge of these structures enable Chinese speakers to interpret the conceptual relationships. The difference between central participants and peripheral participants lies in the fact that the former are positionally marked, while the latter are usually related by prepositions. The distinction between participants and circumstantials is difficult to maintain since individual processes treat them differently. The participants presupposed by a processes are best regarded as prototypes to account for the metaphorical uses and the exclusion of the non-prototypical instances of a category. Participants are also grouped on the basis of the fixedness of their categories. The conceptual relationships also have their hierarchical structure. At the top level, there are P1, P2 and P3. At the bottom, the relationships vary with each individual process. At the intermediate level, some conceptual roles can be established to capture the similarities of relationships. The types of knowledge investigated in this thesis are presented as entries in a conceptual dictionary.

This study investigates grammatical voice from a functional-semantic viewpoint. While previous studies have focussed mainly on the active-passive relationship or at times the active-middle relationship, this study takes a more comprehensive approach, assuming voice in a given language to be a system of values for expressing participant-to-event and subject-to-verb relations. The primary research compares the voice systems of four languages (two Indo-European and two non-Indo-European) in depth in order to discover the overarching motivation for their several organizations. Both inflectional and sentence-derivational voice types are included, with an attempt to integrate their functions. The interrelationships of voice with other grammatical systems such as aspect and modality are examined as well. The study found that expressions of voice have two significant patterns of organization, motivated either pragmatically by considerations of empathy and topicalization (as in English and Hungarian) or semantically by modification of the properties of the subject and/or its relationship to the event (Russian and the affective passives of Japanese and English). These two motivating principles may and often do intersect and overlap, producing the complexity which has proved so formidable in the study of voice. The notion of a 'basic' argument structure of the verb, which has serious implications for any analysis of voice, was explicated and shown to be inapplicable to some languages (e.g. Hungarian); the notion of a given language being 'biased' towards a transitive or an intransitive conceptualization of events was also found to be material to voice organization. Finally, the investigation of the interaction of voice with aspect and modality reveals that these three systems (and perhaps others) cooperate to produce a holistic perspective on an event in terms of actor-orientedness or patient-orientedness.

The present analysis studies aspect from a triple perspective: morphological, semantic and syntactic. This division aims at clarifying the concept of aspect, accounting for the traditional dichotomy Imperfect/Preterit tense in Spanish. In contrast to those authors who consider the above categories as separate entities, we claim that the term aspect should be studied as an interaction of the three categories in the following sequence: (1) Morphological/semantic interaction. This interaction shows that the situation a verb expresses (whether durative, punctual, iterative, etc.) may be influenced by the choice of the Imperfect or the Preterit. (2) Morphological/syntactic interaction. This interaction illustrates that the choice of the Imperfect or the Preterit tenses involves different sentential components, mainly subject and adverbials. (3) Semantic/syntactic interaction. This interaction reveals that the situation a verb designates may be affected by elements of the sentence such as subject, direct and indirect object, adverbials and periphrastic verb forms.

Linguistic information, representing the knowledge that a speaker has of his or her language, can be uniformly represented as a system of signs. In addition to the traditional linguistic signs, the connections between morphemes or morphemic words and their meanings, syntactic constructions, idiomatic expressions, discourse phenomena, all types of linguistic information, are treated as meaningful and therefore capable of being represented as signs. A four-strata version of linguistic structure is adopted, with graphic, morphemic, lexemic, and sememic levels. The signs mediate between the information at the different levels, such that the expression and content of each sign is "local", and expression and content in the more general sense (all the way from graphemes to sememes) are related indirectly. The signs which constitute a core grammar of English are formalized and set forth. This formalization, the major goal of the dissertation, is not simply a description of texts, however. Rather, it is an attempt to construct a cognitive model which accounts for those texts. Two points of theory are: (1) there is a distinction between relations (signs) and processes (encoding and decoding), and (2) language is an adaptive system subject to continuous modification. The linguistic information in the semiotic format is shown to be useable to produce and decipher texts, and to be readily modified when new information is encountered.

Taiwanese has been analyzed either as a suppletive form of le or an equivalent of a higher abstract verb 'YOU', that asserts the existence of an event or a state. The latter seems plausible. However, when we carefully examine the various semantic functions of u, we find that they lack the constancy which would allow them to be united as a higher existential 'U'. This does not mean that there are various separate u's. We propose there is only one morpheme u. The 'major' various semantics for u, i.e. 'expectation', 'emphasis', 'perfectivity' and 'existence-possession', are, metaphorically connected. The links between them are reflected in a series of ambiguous sentences with u, in which an identical form of utterance may represent different meanings according to different contexts of use.

This thesis is a description of basic Chinese grammar. Unlike traditional methods, the descriptions uses the Dynamic Grammar Notation System, a tool designed by Sydney M. Lamb for language description and language processing. All of the syntactic and semantic description is formalized into rules, so the grammar can be used either to understand basic Chinese syntax or as a rule base in natural language processing. Some issues related to machine implementation are also discussed in the first part of the thesis.