This thesis models the importance of reputation in indefinite interactions using experimental and empirical evidence. In modern society reputation plays an important role in many everyday life situations. I analyse the resulting incentive structures through microeconomic modelling and econometric regression analysis. Thus, I gauge the impact of reputational considerations on individual decision making.

The dissertation investigates syntactic and semantic aspects of the indefinite pronoun system in Modern Hebrew and consists of an experimental part and a theoretical part. The experimental part presents the grammaticality judgment task conducted to test three theoretical questions: (1) the relation between negation and the licensing of AF `any' and KOL `any'; (2) the contrast between EYZE `some' and EYZESEHU `some' in terms of specificity; and (3) the correlation between syntactic position and free choice readings. Three main theoretical findings are contributed by this work. First, it is shown that from a typological viewpoint, Hebrew, a Semitic language, patterns with Romance and Germanic languages, rather than Japanese-type languages, in having indefinite pronouns specialized for particular operators available in the discourse. Second, the thesis proposes a novel unified syntax-semantics for KOL which accounts for its interpretational variability. Working with the Kratzer and Shimoyama (2002) framework where indefinite pronouns generate sets of individual alternatives, the semantics of a KOL--modified noun phrase is formalized as a variable and containing a restriction. When this restriction ranges over kinds, KOL receives a generic reading; when the restriction is over a contextually specified set of entities, KOL has an episodic reading. In these cases, the KOL--phrase moves to the argument position of a universal quantifier which binds the individual alternatives generated by the KOL--phrase. If KOL stays in situ, the individual alternatives are allowed to expand into propositional alternatives, resulting in the free choice reading. Third, I discuss the DP-internal structure of [eyze(N)se-hu (N)] and [(N) kol(N)se-hu], treating 'se-hu' as a CP. I propose that there is a correlation between the postnominal position and the free choice readings of these pronouns, suggesting that domain restrictions, usually derived in the semantics-pragmatics, may also be encoded in the syntax. If this hypothesis is on the right track, it could provide us with a better understanding of how and when in the process of language acquisition domain restrictions found with indefinite pronouns are acquired.

The dissertation investigates syntactic and semantic aspects of the indefinite pronoun system in Modern Hebrew and consists of an experimental part and a theoretical part. The experimental part presents the grammaticality judgment task conducted to test three theoretical questions: (1) the relation between negation and the licensing of AF `any' and KOL `any'; (2) the contrast between EYZE `some' and EYZESEHU `some' in terms of specificity; and (3) the correlation between syntactic position and free choice readings. Three main theoretical findings are contributed by this work. First, it is shown that from a typological viewpoint, Hebrew, a Semitic language, patterns with Romance and Germanic languages, rather than Japanese-type languages, in having indefinite pronouns specialized for particular operators available in the discourse. Second, the thesis proposes a novel unified syntax-semantics for KOL which accounts for its interpretational variability. Working with the Kratzer and Shimoyama (2002) framework where indefinite pronouns generate sets of individual alternatives, the semantics of a KOL--modified noun phrase is formalized as a variable and containing a restriction. When this restriction ranges over kinds, KOL receives a generic reading; when the restriction is over a contextually specified set of entities, KOL has an episodic reading. In these cases, the KOL--phrase moves to the argument position of a universal quantifier which binds the individual alternatives generated by the KOL--phrase. If KOL stays in situ, the individual alternatives are allowed to expand into propositional alternatives, resulting in the free choice reading. Third, I discuss the DP-internal structure of [eyze(N)se-hu (N)] and [(N) kol(N)se-hu], treating 'se-hu' as a CP. I propose that there is a correlation between the postnominal position and the free choice readings of these pronouns, suggesting that domain restrictions, usually derived in the semantics-pragmatics, may also be encoded in the syntax. If this hypothesis is on the right track, it could provide us with a better understanding of how and when in the process of language acquisition domain restrictions found with indefinite pronouns are acquired.

Wh-indefinites in Chinese have received great attention and discussion in the literature. This thesis investigates the status of Chinese wh-indefinites and their behaviors in the so-called donkey sentence. A typical example of wh-indefinite will be like the following: (4) Ta bu xiang zai shuo shenme le he not want again say what LE ¡¥He does not want to say anything again.¡¦ ¥L¤£·Q¦A»¡¤°»ò¤F¡C In this thesis, accounts of wh-indefinites as variables, polarity items and quantifiers from Huang (1982), Li (1992), Cheng (1991, 1994) and Lin (1996, 1998) are provided and this thesis discusses the problems each account presents. Countering Li¡¦s (1992) work, this work argue that wh-indefinite like zenmeyang ¡¦how¡¦ and weishenme ¡¥why¡¦ cannot be directly treated as variables since they feature variables only if they appear after an auxiliary (under which situation they will express purpose or method) in a donkey sentence according to Tsai (1999, 2000), see the following examples: (5) a. Akiu hui wei (le) shenme cizhi wo jiu hui wei (le) shenme cizhi¡@ Akiu will for LE what resign I then will for LE what resign ¡¥If Akiu will resign for the purpose x, I will then also resign for the purpose x.¡¦ ªüQ·|¬°¡]¤F¡^¤°»òÃ㾡A§Ú´N·|¬°¡]¤F¡^¤°»òÃ㾡C b. *Akiu wei (le) shenme hui cizhi wo jiu wei (le) shenme hui cizhi Akiu for LE what will resign I then will LE what will resign Intended ¡¥If Akiu will resign because of the reason x, I will then also resign because of the reason x. *ªüQ¬°¤F¤°»ò·|Ã㾡A§Ú´N¬°¤F¤°»ò·|Ã㾡C (Tsai¡¦s 2000, 15 glossed and translated by the author) Avoiding treating wh-indefinites as variables, we may treat them as quantifiers according to Huang¡¦s observation. But Huang¡¦s (1982) quantifier account of wh-indefinites is also rejected because I observe that wh-indefinites do not exhibit island sensitivity. Then we are led to another treatment: treating wh-indefinites as polarity items because they are sensitive to polarity environment. Aside from the sensitivity to polarity environment, Lin¡¦s (1996) work argues that wh-indefinites are licensed as polarity items if the sentence is subject to NEEC- non-entailment of existence condition. NEEC tells us that if (part of) a sentence does not have the existential import of object; wh-words can be licensed as wh-indefinites. This faces challenge because there are cases presupposing existential import yet the wh-indefinite is licensed. In chapter 3, two types of donkey sentences are identified in Cheng and Huang¡¦s (1996) work and are accounted with different approaches-Unselective Binding and E-type analysis. Though accepting their viewpoint generally, Lin (1996) makes a distinction between one-case and multi-case reading and opens the possibility of universal interpretation in ruguo-conditionals. Later I introduce Indirect binding approach to account both types of donkey sentences. Indirect Binding argues that some quantified expression plays the role as associating the indefinite with the anaphoric element in donkey sentences: (6) [Everyone who ti keeps a dogj]i like itj. The example here captures the very basic idea of Indirect Binding: the chunk of expression, which has the indefinite a dog in its scope, c-commands the pronoun it and hence relates it to the indefinite. I extend it to the analysis of Chinese donkey sentences and discover that although Indirect Binding does not target at a particular status, the condition that the indirect binder must c-command the indefinite seems to suggest that c-commanding is the way of licensing a wh-indefinite, which echoes Li¡¦s and Cheng¡¦s analyses. In Chapter 4, I introduce wh-indefinites in Japanese and Korean respectively and conclude that quantificational force comes from the environment but not from wh-indefinites themselves, countering Huang (1982). Finally, I suggest that licensing a wh-word as a wh-indefinite is through being c-commanded by the relevant operators. It is so because of Li¡¦s and Cheng¡¦s observation of wh-indefinites licensing and also of a condition in Indirect Binding that indirect binders must c-command the indefinite in order to have it as in its scope, as I have mentioned. This thesis then provides a viewpoint that perhaps, wh-indefinites can be treated uniformly as polarity items considering that Indirect Binding approach explains their behavior in donkey sentences. This thesis also opens the door for analyzing wh-indefinites in donkey sentences under Indirect Binding in other languages. Keywords¡G Wh-indefinite, polarity item, variable, quantifier, donkey phenomenon, Indirect Binding

The dissertation investigates syntactic and semantic aspects of the indefinite pronoun system in Modern Hebrew and consists of an experimental part and a theoretical part. The experimental part presents the grammaticality judgment task conducted to test three theoretical questions: (1) the relation between negation and the licensing of AF `any' and KOL `any'; (2) the contrast between EYZE `some' and EYZESEHU `some' in terms of specificity; and (3) the correlation between syntactic position and free choice readings. Three main theoretical findings are contributed by this work. First, it is shown that from a typological viewpoint, Hebrew, a Semitic language, patterns with Romance and Germanic languages, rather than Japanese-type languages, in having indefinite pronouns specialized for particular operators available in the discourse. Second, the thesis proposes a novel unified syntax-semantics for KOL which accounts for its interpretational variability. Working with the Kratzer and Shimoyama (2002) framework where indefinite pronouns generate sets of individual alternatives, the semantics of a KOL--modified noun phrase is formalized as a variable and containing a restriction. When this restriction ranges over kinds, KOL receives a generic reading; when the restriction is over a contextually specified set of entities, KOL has an episodic reading. In these cases, the KOL--phrase moves to the argument position of a universal quantifier which binds the individual alternatives generated by the KOL--phrase. If KOL stays in situ, the individual alternatives are allowed to expand into propositional alternatives, resulting in the free choice reading. Third, I discuss the DP-internal structure of [eyze(N)se-hu (N)] and [(N) kol(N)se-hu], treating 'se-hu' as a CP. I propose that there is a correlation between the postnominal position and the free choice readings of these pronouns, suggesting that domain restrictions, usually derived in the semantics-pragmatics, may also be encoded in the syntax. If this hypothesis is on the right track, it could provide us with a better understanding of how and when in the process of language acquisition domain restrictions found with indefinite pronouns are acquired.

The dissertation investigates syntactic and semantic aspects of the indefinite pronoun system in Modern Hebrew and consists of an experimental part and a theoretical part. The experimental part presents the grammaticality judgment task conducted to test three theoretical questions: (1) the relation between negation and the licensing of AF `any' and KOL `any'; (2) the contrast between EYZE `some' and EYZESEHU `some' in terms of specificity; and (3) the correlation between syntactic position and free choice readings. Three main theoretical findings are contributed by this work. First, it is shown that from a typological viewpoint, Hebrew, a Semitic language, patterns with Romance and Germanic languages, rather than Japanese-type languages, in having indefinite pronouns specialized for particular operators available in the discourse. Second, the thesis proposes a novel unified syntax-semantics for KOL which accounts for its interpretational variability. Working with the Kratzer and Shimoyama (2002) framework where indefinite pronouns generate sets of individual alternatives, the semantics of a KOL--modified noun phrase is formalized as a variable and containing a restriction. When this restriction ranges over kinds, KOL receives a generic reading; when the restriction is over a contextually specified set of entities, KOL has an episodic reading. In these cases, the KOL--phrase moves to the argument position of a universal quantifier which binds the individual alternatives generated by the KOL--phrase. If KOL stays in situ, the individual alternatives are allowed to expand into propositional alternatives, resulting in the free choice reading. Third, I discuss the DP-internal structure of [eyze(N)se-hu (N)] and [(N) kol(N)se-hu], treating 'se-hu' as a CP. I propose that there is a correlation between the postnominal position and the free choice readings of these pronouns, suggesting that domain restrictions, usually derived in the semantics-pragmatics, may also be encoded in the syntax. If this hypothesis is on the right track, it could provide us with a better understanding of how and when in the process of language acquisition domain restrictions found with indefinite pronouns are acquired.

No description available.

Cette thèse porte sur les particularités de la reprise anaphorique des pronoms indéfinis et des noms épicènes qui ont pour référents des individus humains indéfinis. Nous allons essayer de distinguer les différents facteurs (morphosyntaxiques, sémantiques et pragmatiques), qui influent sur le choix du pronom anaphorique. Ce choix ne sera pas toujours objectif. Il ne sera pas obligatoirement conforme aux règles d’accord syntaxique, puisqu’il dépendra de la décision de l’énonciateur concernant la pertinence du sexe ou du nombre d’individus faisant l’objet du discours. Ainsi, il est prouvé que le pronom personnel de la troisième personne du pluriel they est le meilleur candidat que ce soit dans les contextes où l’énonciateur et ses co-énonciateurs ignorent le sexe et/ou le nombre des personnes évoquées, dans les contextes génériques ou bien dans les contextes où l’énonciateur ne souhaite pas être précis sur l’identité de la personne dont il ou elle parle. Les pronoms personnels de la troisième personne du singulier he / she ne sont pas préférés dans ces contextes puisqu’ils rendent le référent plus défini qu’il l’est dans le discours en ajoutant des propriétés telles que le sexe et le nombre. Le choix que l’on fait pour se référer ou se coréférer sous-entend la prise d’un point de vue vis-à-vis des référents ou vis-à-vis des interlocuteurs. L’étude de plusieurs corpus, écrits comme oraux (BNC, COCA, SOAP, CHILDES, le corpus littéraire, le corpus de presse et d’émissions de télévision), nous a dévoilé que les locuteurs anglophones savent en quelque sorte manipuler le système pronominal conventionnel afin de rendre explicite ou de mettre en avant certains éléments ou bien d’en sous-spécifier ou d’en taire d’autres. / This dissertation investigates the peculiarities of pronominal reference to indefinite pronouns and epicene nouns that refer to indefinite human beings. It brings to light morphosyntactic, semantic and pragmatic factors which influence the choice of the anaphoric pronoun. This choice is not always objective, and does not necessarily match the rules of syntactic agreement. It tends to depend on the speaker’s decisions about the relevance of the sex and/or the number of individuals that are the topic of discussion. Thus, the third person plural pronoun they proves to be the best candidate when the referent’s sex and/or number is unknown by both the speaker and the hearer, as well as in generic contexts or when the speaker wishes to remain vague about the identity of the person he or she is referring to. The third person singular pronouns he / she do not fit in these contexts, since they render the referent more definite than it actually is in the discourse by adding such properties as sex and number. The choices that we make when referring or coreferring reveal our attitudes towards either referents or addressees. The study of both written and oral corpora (BNC, COCA, SOAP, CHILDES, literary, press and talk show corpora) reveals how English speakers are able to manipulate the conventional pronominal system so as to render explicit or highlight certain attributes and to underspecify others.

Please download this dissertation to see the abstract.

Active learning is a subfield of machine learning that has been successfully used in many applications. One of the main branches of active learning is query synthe- sis, where the learning agent constructs artificial queries from scratch in order to reveal sensitive information about the underlying decision boundary. It has found applications in areas, such as adversarial reverse engineering, automated science, and computational chemistry. Nevertheless, the existing literature on membership query synthesis has, generally, focused on finite concept classes or toy problems, with a limited extension to real-world applications. In this thesis, I develop two spectral algorithms for learning halfspaces via query synthesis. The first algorithm is a maximum-determinant convex optimization method while the second algorithm is a Markovian method that relies on Khachiyan’s classical update formulas for solving linear programs. The general theme of these methods is to construct an ellipsoidal approximation of the version space and to synthesize queries, afterward, via spectral decomposition. Moreover, I also describe how these algorithms can be extended to other settings as well, such as pool-based active learning. Having demonstrated that halfspaces can be learned quite efficiently via query synthesis, the second part of this thesis proposes strategies for mitigating the risk of reverse engineering in adversarial environments. One approach that can be used to render query synthesis algorithms ineffective is to implement a randomized response. In this thesis, I propose a semidefinite program (SDP) for learning a distribution of classifiers, subject to the constraint that any individual classifier picked at random from this distributions provides reliable predictions with a high probability. This algorithm is, then, justified both theoretically and empirically. A second approach is to use a non-parametric classification method, such as similarity-based classification. In this thesis, I argue that learning via the empirical kernel maps, also commonly referred to as 1-norm Support Vector Machine (SVM) or Linear Programming (LP) SVM, is the best method for handling indefinite similarities. The advantages of this method are established both theoretically and empirically.