Natural Language Processing (NLP) systems usually require large amounts of pre-coded domain knowledge to perform semantic analysis automatically. Until repositories of such background knowledge are widely available, these systems may not scale up to non-trivial applications of NLP. This paper describes the design and implementation of a system that uses surface-syntactic information to interpret interactively semantic relationships between clauses. English technical texts are analyzed by a domain-independent parser that produces detailed parse trees of the input. The system then examines clausal connectives and syntactic verb phrase features to determine what kinds of semantic relationships exist between clauses. The results of this activity are used in a large Knowledge Acquisition system that, by design, requires little a priori semantic knowledge. We present a set of semantic labels appropriate to syntactically connected clauses (Clause-Level Relationships) and a description of the...

The set of English modal verbs is widely recognized to communicate two broad clusters of meanings: epistemic and root modal meanings. A number of researchers have claimed that root meanings are acquired earlier than epistemic ones; this claim has subsequently been employed in the linguistics literature as an argument for the position that English modal verbs are polysemous (Sweetser, 1990). In this paper I offer an alternative explanation for the later emergence of epistemic interpretations by linking them to the development of the child's theory of mind (Wellman, 1990); if correct, this hypothesis might have important implications for the shape of the semantics of modal verbs.

This thesis proposes new analyses of English missing object constructions (mocs) (e.g. the tough construction, purpose infinitives, etc.) and parasitic gap formation. These analyses are formulated in the framework of Head-Driven Phrase Structure Grammar (hpsg). hpsg divides unbounded dependency constructions (udcs) into two classes depending on whether the filler is in argument or non-argument position. mocs have argument fillers and are classified as weak udcs. The evidence that motivates the weak udc analysis is re-evaluated and it is claimed that, in fact, mocs are not udcs. It is proposed that a lexical rule promotes missing objects from the comps to the subj list in much the same way as passive promotes objects. In contrast to passive, the original subject is not demoted and missing object vps have two elements in subj, both available to be controlled. Raising and Equi signs are modified to permit them to inherit second subj members from their complements: in this way the appare...

Knowledge acquisition from text is often attempted in the presence of large amounts of pre-coded domain knowledge. Seeding a system with such knowledge is often a huge knowledge acquisition effort in itself. Breaking this circle requires a text processing system that relies on little a priori semantic information. In the absence of such knowledge, processing must rely on available information, such as surface syntax. Semantic analysis will consist of determining the semantic relationships between the various surface-syntactic constituents in the sentence. This report describes the design and implementation of a system that interactively interprets semantic relationships between clauses. The Clause-Level Relationship Analyzer inspects the parse trees produced by a domainindependent parser for syntactic features that can be used to determine what kinds of semantic relationships exist between clauses. The system is designed to require as little a priori semantic knowledge as possible. This document offers a set of semantic labels appropriate to syntactically connected clauses and a description of the theory behind assigning these labels to particular inputs. It also presents elements of the implementation of the Clause-Level Relationship Analyzer and looks at its performance. 1

Abstract: The semantics of the linguistic modals is argued to be determined mainly by the power structure of the participants in the interaction. In the deontic uses of the modals, another determining factor is the expectations of the participants ’ attitudes towards the relevant action. By viewing the evidence as a power in its own right, our analysis can be generalized to the epistemic uses in a coherent way. The epistemic uses are seen as pragmatic strengthenings of the deontic uses, rather than as metaphorical mappings.

This paper presents ‘source’ as the distinctive feature for a twofold semantic categorization for Chinese modal expressions. Previous studies have characterized Chinese modals as words used to express the speaker’s opinion or attitude. Yet given the absence of morphological and syntactic distinctions in Chinese, there has been little consensus among different accounts as to within what limit this definition is to apply. Instead of imposing such preconceived cross-linguistic perspective, this paper argues that the component of ‘source’ must be taken into consideration to outline a clearly specified semantic category in Chinese modal systems. A distinction is drawn between modals with ‘the source of opinion or attitude ’ as part of their meaning components and those without. The former is non-subject-oriented by nature, including modals traditionally seen as auxiliaries (e.g., epistemic keneng ‘may’, and deontic keyi ‘can’) and adverbs (e.g., epistemic yiding ‘must’, deontic wubi ‘must’, and evaluative xingkui ‘fortunately’). The latter type, being subject-oriented, functions as the main verb in a sentence (e.g., epistemic caice ‘guess’, deontic yaoqiu ‘demand’, and evaluative qingxing ‘be gratified’) and covers a group of words that have been widely identified as auxiliaries (e.g., dynamic neng ‘can’). ‘Neutral possibility’ as advanced by a number of scholars is also proven to belong to dynamic modality because it takes the enabling condition as its subject and is subject-oriented in the sense that the proposition it qualifies concerns the capacity of its subject. The source involvement property alongside the bipartite model provides a unified account for Chinese modal inventories. It entails formal dimensions such as argument selection and categorical manifestation and also reflects the speaker’s motivation in exploiting varied modal types to achieve different pragmatic purposes.

CONTENTS Introduction. The expression of modality in Korean i I. Scope of investigation i II. Structure of thesis i III. General expository practice ii III.I. Abbreviations ii III.II. Data transcription iii III.III. Data glossing iv III.IV. Data sources and source abbreviations v Chapter 1 Preliminaries 1 1.1. Modality 1 1.1.1. The concept of modality 1 1.1.2. A notional model of modality 5 1.1.3. A taxonomy of modalities 13 1.1.3.1. The basic dichotomy - epistemic and deontic modality 14 1.1.3.2. Epistemic modality 15 1.1.3.3. Deontic modality 17 1.1.3.4. Evidentials 19 1.1.3.5. Dynamic modality 20 1.1.3.6. Volitive modality 22 1.1.3.7. Other types of modality 23 1.2. Korean 24 1.2.1. General language data 24 1.2.2. Classification 25 1.2.3. Phonology 27 1.2.4. Morphology 29 1.2.5. Syntax 31 Chapter 2 Mood 34 2.1. The concept of mood 34 2.2. The verbal system of speech levels 39 2.3. A taxonomy of mood in Korean 42 2.3.1. Neutral mood 2

In an empirical study participants were asked to rate the perceived degree of certainty of utterances that contained either the modal would or main verb be (e.g. That would be me vs. That’s me), and which were also variously produced with one of three intonational contours (downstepped, declarative, and yes-noquestion). We found that both downstepped contour and epistemic would made a significant and independent contribution to the assessment of speaker certainty. That is, participants rated utterances with the downstepped contour as most certain, followed by those with the declarative contour, while the yes-noquestion contour was perceived as highly uncertain. Similarly, participants rated speakers ’ responses with epistemic would as significantly more certain than those without it. 1.

The determinate features of tense, mood and modality, which are interwoven in the conditional construction form a well-known problem for L2 learners; however, the factors impede acquisition have never been satisfactorily treated. This paper attempts to explore the L2 acquisition of if-conditionals, by employing error analysis based on a form-function mapping framework. Disengaging the expression of conditionality into two dimensions: temporality and hypotheticality, we identified the features of the forms and functions of different conditional types, and created a new scheme to account for their mapping, on which the acquisition stages were inferred. It was argued that the heavy load of content is less amenable to adequate production than the complexity of lexical shape. The data was obtained from CLEC (Chinese Learner English Corpus), a big-scale corpus consisting of roughly 1,000,000 words with error tags. The results showed: (1) the acquisition sequence was parallel to our prediction on the form-function mapping underpinnings (2) the misuses could be ascribed to mapping failure. Additionally, this present work serves to provide some explanatory accounts, in the hope of unveiling the mystery of the arduousness of acquiring conditionals, and contribute to the tightening of acquisition theorizing.

Propositional attitudes represent a highly debated topic within the philosophical semantic tradition of research, as well as within the truthconditionally oriented branch of semantics, and they have persistently provided very lively and problematic areas for discussion.