This dissertation is on the semantics of modal expressions and attitude verbs like English ‘must,’ ‘may,’ ‘ought,’ and ‘want’ — expressions that allow us to discuss socially important modal facts like obligations, permissions, and desires. When discussing them, we rarely spell out the modal facts with complete specificity. For example, I may say, ‘I want ice cream,’ without specifying that what I desire, in particular, is unmelted, non-toxic, chocolate ice cream. If, in response, someone were to give me melted, toxic, vanilla ice cream, it would be fair for me to reply that what I have been given is not what I wanted. Similarly, I can truly say ‘you must wash the dishes,’ or ‘you may have some wine,’ without specifying that there are exceptions, that is, specific ways of washing the dishes, or of having wine, that are not ways of doing what you must or are allowed. I call modal claims that have such exceptions underspecific. Since avoiding underspecificity may require explicitly ruling out an infinite number of exceptions, most of our ordinary modal claims are underspecific. Our reliance on them is thus crucial to our very capacity to communicate about the modal facts. I argue that modal claims can be true and underspecific by virtue of the meanings of the modal expressions that figure in them, and I defend a general semantic framework for modals that explains how. In the simplest cases, my framework predicts that speakers may describe the contents of desires, permissions, and obligations using conditions that are merely necessary for their fulfillment. Since getting unmelted, non-toxic chocolate ice cream necessitates getting ice cream simpliciter, this explains how I may truly state ‘I want ice cream’ even if the content of my desire is more specific. Generalizing this account, however, poses significant challenges. Indeed, many philosophers and linguists have rejected semantic theories of modals that allow for true, underspecific modal claims for two main reasons. One is that previous accounts give rise to several well-known logical puzzles, including the puzzle of free choice permission, Ross’s puzzle, and the Samaritan paradox. A second reason is that existing accounts generate the wrong truth value judgments in examples involving complex preferences and forms of uncertainty. In response, theorists have offered semantics for modals like ‘ought’ and ‘want’ that rely on the theory of rational choice and the notion of expected utility. Straightforward expected utility analyses succeed in generating the desired truth value judgments, but at the cost of giving up the possibility of underspecifying the modal facts in the ways that we manifestly do. I argue that we can meet both of these challenges, while still allowing for true, underspecific modal claims. In order to do so, I develop new solutions to the puzzle of free choice permission, Ross’s Puzzle, and the Samaritan paradox that improve on the empirical predictions of existing solutions. My framework also generates a methodologically desirable degree of independence between the proper semantics of modal expressions, on the one hand, and the proper analyses of desire or obligation in terms of rational choice, on the other. As a result, the semantic framework I develop in this dissertation can succeed where theorists who have turned to decision-theoretic semantics for ‘want’ and ‘ought’ have thought that existing theories fail. Besides the advantages that my semantic framework gains with respect to these problems in particular, it more generally provides new insights into the rich linguistic capacities and conventions that speakers leverage in order to communicate about the modal facts.

In this thesis, I develop a lexicon-syntactic study about the internal structure of the personal pronouns in Brazilian Portuguese (henceforth BP), aiming the investigation of which the formative features of a pronoun in BP are and their relationship with the syncretism phenomenon. Empirically, I show a description of the internal structure of the personal pronouns in BP, showing that the traditional φ-features which build a pronoun (person, number and gender) are actually categories which bare more elemental features which define the content and the shape of a pronoun. More elementary component structures of the categories person, number and gender are able to describe satisfactorily the pronoun paradigm in BP. Therefore, the different pronouns (and their syntactic roles) can be described through their inner composition, holding some features, such as [SPECIFIC], once considered out of their structure. Theoretically, I define which elementary formative features that form a pronoun are, how this composition is made and which its syntactic consequences are. To do so, I adopt a feature geometry developed based on Harley & Ritter (2002) and Béjar (2003)`s proposals. I assume, then, that the pronoun formative features obey a hierarchy which is based on underspecification. The φ-theory proposed by Béjar (2003; 2008) supports adequately the pronoun compositionality developed for BP. Likewise, decomposition for Case categories is made as an optimal solution for Case configuration, which takes into account evidence from languages like English and BP, which present Case differences only in their personal pronouns. Case, then, is treated geometrically, analogue to that proposed to φ-features. As a result, Case categories in BP obey the geometry [C[OBL[GEN][ABL]]]. A value mechanism for these features is proposed also analogue to the one for φ-features. / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Nesta tese, desenvolvo um estudo lexico-sintático sobre a estrutura interna dos pronomes pessoais em Português Brasileiro (doravante PB), visando investigar quais são os traços formativos do pronome em PB e sua relação com o fenômeno do sincretismo. Empiricamente, apresento uma descrição da estrutura interna dos pronomes pessoais em PB, mostrando que os tradicionais traços φ (pessoa, número e gênero) componentes destes pronomes são na verdade elementos categoriais que comportam traços mais elementares os quais definem tanto o conteúdo quanto a forma do pronome. Estruturas componentes mais atômicas das categorias pessoa, número e gênero são capazes de descrever satisfatóriamente o paradigma pronominal em PB. Assim, os diferentes pronomes (e os papéis que eles desempenham na sintaxe) podem ser descritos através de sua composição interna, compreendendo alguns traços, como [SPECIFIC], por exemplo, outrora considerados externos ao conjunto de traços φ. Teoricamente, defino quais são estes traços formativos mais elementares que compõem o pronome, como esta composição é feita e quais são seus reflexos sintáticos. Para tal, adoto uma geometria de traços desenvolvida com bases nas propostas de Harley & Ritter (2002) e Béjar (2003). Assumo, portanto, que os traços formativos do pronome seguem uma hierarquia e esta se dá com bases na subespecificação. A proposta de uma teoria φ, de Béjar (2003; 2008), suporta adequadamente a composicionalidade pronominal desenvolvida para o PB. Da mesma maneira, uma decomposição das categorias de Caso é feita como uma solução ótima para a configuração de Caso, levando em conta principalmente evidências de línguas como o inglês e o PB, que apresentam diferenças Casuias apenas em seus pronomes pessoais. Caso, portanto, é tratado a partir de uma geometria de traços, análoga àquela proposta para os trços φ. Assim, as categorias de Caso para o PB obedecem a geometria [C[OBL[GEN][ABL]]]. Um mecanismo de valoração destes traços é proposto também análogo àquele proposto para os traços φ.