Meaning seems at once the most obvious and the most obscure aspect of language to be studied. Since most lexical items have several meanings the rules which combine them into sentences will frequently yield several possibilities for interpretation. Potential ambiguity is usually resolved unconsciously in speakers' minds, but may occasionally cause an obstacle in communication, and has therefore been considered a deficiency of language. One of the goals of linguistic research is to illuminate the processes which occur in speakers' minds by studying the organisational structure of concepts and the interrelations between them. The aim of this study is to describe the structural properties of lexical items with multiple meanings, in particular polysemous lexemes, by means of three case studies in English, German and Turkish. The first case study explores paradigmatic sense relations of identity and inclusion (vertical relations). The study involves structural comparison between the nets of each of the senses of a polysemous lexeme in English in contrast with its corresponding forms and senses in German and Turkish. Findings suggest that communicative problems can arise due to intra-domain specific ambiguity. The second case study involves the structural description of a polysemous lexeme in German. Theories of paradigmatic sense relations of opposition and exclusion (horizontal relations) are used to investigate sense opposition at the micro-level which is a specific phenomenon in polysemy. Unlike the first case study, collocational patterns are focused on by looking into syntagmatic sense relations. Results show that theories of sense opposition at the macro-level are applicable to sense opposition at the micro-level. The third case study also deals with the structural description of a polysemous lexeme, this time in Turkish. Besides the description of some disambiguation processes, the study demonstrates that sense opposition at the micro-level exists in Turkish, and that relevant theories are applicable just as in German.

In this thesis I examine the collocational behaviour of semantically-related verbs (e.g. head, run, manage) and nouns (e.g. issue, factor, aspect) from the domain of business English. The results of this corpus-based study show that much of the collocational behaviour exhibited by these lexical items can be explained by examining a number of linguistic features and processes which influence the process of collocation. These include the semantics of the individual lexical items being studied, the use of metaphor, semantic prosody, and the phraseological nature of some of the lexical items. I show that it is possible to explain, and therefore teach, certain collocations by considering the linguistic features and processes which influenced the way they were formed. My contention is that if the learner is encouraged to look for an explanation, it makes the process of learning collocations more meaningful and, as a result, more memorable.

When David Kaplan put forward his theory of sentence truth incorporating demonstratives, initially proposed in ‘Dthat' (1978) and later developed in ‘Demonstratives' (1989a) and ‘Afterthoughts' (1989b), it was, to his mind, simply a matter of book-keeping, a job that had been pushed aside as a complication when a truth conditional semantics had been proposed. The challenges considered in this thesis are challenges to the effect that Kaplan's theory of sentence truth is, for one reason or another, inadequate. My overarching aim is to defend Kaplan's theory of sentence truth against these challenges. In chapter one I am concerned only with setting out some preliminary considerations. In chapter two I defend Kaplan's theory of sentence truth against a general challenge, motivated by linguistic data from ‘contextualists' and ‘relativists'. I argue that the methods and data employed by proponents of contextualism and relativism are lacking and as such should not be taken to have seriously challenged Kaplan's theory of sentence truth. In chapter three I defend Kaplan's theory of sentence truth against challenges to the effect that his theory is not suited to delivering on its initial purpose—to provide a semantics for indexical and demonstrative terms. I then develop a form of semantic pluralism that I take to be entirely compatible with the Kaplanian model. In chapters four I demonstrate the efficiency of this Kaplanian model when it comes to defending Kaplan's theory against the challenge of providing suitable semantics to accommodate discourse involving future contingents. And finally, in chapter five I consider contextualist accounts of discourse concerning vague predicates.

No description available.

No description available.

No description available.

No description available.

No description available.

Thai is a discourse pro-drop language (also called "radical pro-drop" language), since it exhibits highly frequent use of null pronouns without the involvement of agreement morphology. The descriptive goal of this thesis is to describe the different syntactic contexts in Thai where null pronouns occur and where they do not occur. The theoretical goal is to explain why pronouns sometimes have to be pronounced, sometimes may, but need not be, and sometimes cannot be pronounced, in Thai, and how this relates to pro-drop as found in other languages. The thesis will thereby hopefully contribute to the theory of pronouns, pronominal reference, and pro-drop. After an introduction (Chapter 1), the distribution of null/overt pronouns and the restrictions on their occurrences are discussed (Chapter 2). It is demonstrated that even though a null pronoun can appear in any argument position, it cannot take the position of a prepositional complement or occur in a conjoined NPs construction. It is also demonstrated that a null pronoun/argument in Thai always looks for an antecedent for a referential reading (Chapter 5). If there is no controlling linguistic antecedent for a null pronoun in a higher clause, then a discourse topic will be the antecedent. If there is no discourse topic either, the speaker is always available as a local antecedent of the null pronoun. The default referential reading of a null pronoun is therefore first person singular. It appears that this generalisation holds true in other discourse pro-drop languages as well (Chapter 3). Correspondingly, a pronoun without an antecedent must be overt. Pronouns with a generic or arbitrary reading are a special case of antecedentless pronouns, which can be null but only when they have generic inclusive reading (Chapter 4). I propose that null pronouns in Thai, and discourse pro-drop languages generally, have no ϕ-features, except for an unvalued referential feature [uR] and a general nominal feature [N]. These features are sufficient for the pronoun to function as an argument, being assigned a θ-role. They are dependent on being bound or controlled by one of the following: (i) a locally c-commanding referential NP, (ii) a null topic operator, which itself is linked to a referential NP in the discourse, (iii) "the speaker", as default, (iv) a generic operator, or (v) a higher generic argument (Chapter 6). That an inclusive generic pronoun is not pronounced (as opposed to other generic/ arbitrary pronouns, which are pronounced) is explained by the presence of a generic operator. The operator behaves just as an adverb that quantifies over the null arguments, i.e. "It is generally true for x." The [uR] feature of the pronoun is probed by this generic operator. As the pronoun has no ϕ-features, it gets a referentially unrestricted reading. This is the inclusive reading of a null generic pronoun. It includes the speaker, the The Syntax of Pro-drop in Thai addressee, and any other people. In other words, the generality of the inclusive reading is from the ϕ-featureless pronoun, in which case it has no restricted reference. Since exclusive generic/arbitrary and quasi-inclusive generic pronouns have more restricted reference, i.e. third person plural and first person plural, respectively, they must be overt when bound by the generic operator. If they were null, they would be indistinguishable from the unrestricted inclusive generic pronoun and the referential first person pronoun "I", since they have no antecedent providing them with features. This means Thai can have relatively unrestricted use of referential third person/impersonal null subjects and a null inclusive generic subject pronoun. I show that this pattern is also found in other discourse pro-drop languages, and is restricted to languages where agreement is not part of sentential syntax. A null argument of the type [uR, N] in turn constitutes a new category in the typology of null arguments. It explains why it cannot function as a prepositional complement, which requires a complement with ϕ-features. This implies that referential null pronominal arguments do not inherit any ϕ-features from their antecedents. The only thing they inherit is the referential index. To summarise, null pronominal arguments do not have ϕ-features, and thus are not pronounced. Pronominal arguments with ϕ-features can be pronounced. In this, and in several other respects, null pronouns in Thai and other discourse pro-drop languages are similar to PRO in languages like English.

This thesis shows a close syntactic relation between yes-no questions (YNQs) and answers (also called yes-no replies, YNRs) in Thai, based on the theory of questions and answers in Holmberg (2010, to appear). To show this correspondence, the semantics and syntax of YNQ particles in Thai are analysed. It is assumed that every YNQ particle in Thai necessarily includes either overt or covert r ‘Q/ or’, a disjunctive particle. As part of a question particle, r ‘Q/ or’ is argued to have the features [Alt(ernative)] and [uFoc]. The [Alt] feature restricts r ‘Q/ or’ to conjoining (or ‘disjoining’) two polarity phrases (PolPs) with identical content but opposite polarity, affirmative or negative. The Pol head of PolP can only merge with verbal categories; therefore, r ‘Q/ or’ conjoins verbal categories only. The [uFoc] feature makes r ‘Q/ or’ the question focus, distinguishing it from a declarative disjunctive sentence with r ‘or’. With these features, YNQs in Thai are seen as disjunctive constructions where r ‘Q/ or’ conjoins two PolPs to form a question of which the second conjunct is deleted at PF. Based on the syntax of the question they mark, YNQ particles are classified into two types. However, particles in both types are derived by the incorporation of the Pol head (and an Adv in certain cases) with the conjunction r ‘Q/ or’, followed by PolP-ellipsis. YNRs in Thai take many different forms and are categorised into primary and secondary answers. Primary YNRs are based on a verb or verb complex from the YNQ (Type-1 questions) or on the question particle itself (Type-2 questions). Secondary YNRs are made up of externally merged materials, typically a particle or particle complex. Following the theory of questions and answers in Holmberg (2010, to appear), these YNRs are assumed to be the carriers of the focused polarity. YNQs have, as an essential component, a variable, which is the polarity, unvalued in the question, and restricted to two possible values: affirmative or negative. This variable is focused in the question. Direct questions ask the addressee to provide a value for this focused, unvalued polarity such that it yields a true proposition. Even minimal YNRs consisting of just one word are full sentential expressions, with an IP which is identical to that of the question, except for the value of the polarity variable, and which is therefore typically not spelled out. To derive primary YNRs to Type-1 questions, the Pol head at Spec, FocP copies the values of the Pol head of one PolP conjunct. This includes a copy of the [V] feature inherited from the verbal complement of the Pol head. The consequence is the elimination of the other conjunct, followed by the spell-out of the copied Pol head at Spec, FocP and deletion of the IP. All that is spelled out, therefore, is a verb or verbal complex ultimately derived from the question, or a negated verb/ verbal complex. This derivation is also applicable to Type-2 questions with the exception that the copied Pol head derives from the question particle itself. Regarding the secondary YNRs, they do not differentiate between two types since they derive from external materials. They are derived by merging a Pol head with an inherent polarity value at Spec, FocP. It can be spelled out as, for example, an honorific particle, an exclamation, a negative word or a polarity particle.