We can think about ourselves in a variety of ways, but only some of the thoughts that we entertain about ourselves will be thoughts which we know concern ourselves. I call these first-person thoughts, and the component of such thoughts that picks out the object about which one is thinking—oneself—the self-concept. In this thesis I am concerned with providing an account of the content of the self-concept. The challenge is to provide an account that meets two conditions on first-person thought. The account must show how we are aware of ourselves when we entertain first-person thoughts, so that we have an account that establishes the cognitive significance of first-person thoughts. But, in addition, this awareness must be as robust as the thinker’s ability to entertain first-person thoughts if our account is to respect the guaranteed referential success of the self-concept. I introduce both the subject matter of the thesis, and the constraints on a satisfactory account of that subject matter in the first chapter. In the second chapter I then set up a further problem: much of our self-knowledge is knowledge of our current mental states and it is often argued that we know about and can ascribe those mental states on the basis of introspection alone. The first constraint on an account of first-person thought described in the preceding paragraph requires that we be aware of ourselves in some way if our thoughts are to have the special cognitive significance of first-person thoughts. Yet, I argue, we neither do nor can introspectively observe a subject of thought and experience when we come to know about our mental states and experiences. The failure of introspection to supply us with perceptual information about a subject of thought presents us with the further potential problem. According to Fregean semantics sense determines reference: we count on the content of the elements of thought to determine the reference of terms that are used to express those elements. If we do not introspectively observe a subject of thought then we seem to be at a loss to account for the concept and we are at risk of having to accept that neither the self-concept nor the first-person pronoun are referential. In the remainder of my thesis I consider various responses that we can offer to this problem. First, I examine whether we can avoid the problem with an alternative account of first-person reference according to which reference is fixed by a reflexive rule, and whether we can also base an account of first-person thought on this account of first-person reference. Secondly, I look at the descriptivist view of first-person thought which could potentially provide both an account of first-person thought and first-person reference. These two suggestions must be rejected on the grounds that they fail to accommodate the special cognitive significance of first-person thought. A third approach to first-person thought argues that we employ an objective self-concept when we think about ourselves, a concept that is informed by bodily experience, rather than by introspective observation of a subject. Yet such an account cannot make sense of first-person thoughts in which we question our own embodiment. Lastly I consider whether it is possible to explain the cognitive significance of first-person thought in terms of non-conceptual first-person contents.

Doctor of Philosophy / Successful communication hinges on keeping track of who and what we are talking about. For this reason, person reference sits at the heart of the social sciences. Referring to persons is an interactional process where information is transferred from current speakers to the recipients of their talk. This dissertation concerns itself with the work that is achieved through this transfer of information. The interactional approach adopted is one that combines the “micro” of conversation analysis with the “macro” of genealogically grounded anthropological linguistics. Murriny Patha, a non-Pama-Nyungan language spoken in the north of Australia, is a highly complex polysynthetic language with kinship categories that are grammaticalized as verbal inflections. For referring to persons, as well as names, nicknames, kinterms, minimal descriptions and free pronouns, Murriny Patha speakers make extensive use of pronominal reference markers embedded within polysynthetic verbs. Murriny Patha does not have a formal “mother-in-law” register. There are however numerous taboos on naming kin in avoidance relationships, and on naming and their namesakes. Similarly, there are also taboos on naming the deceased and on naming their namesakes. As a result, for every speaker there is a multitude of people whose names should be avoided. At any one time, speakers of the language have a range of referential options. Speakers’ decisions about which category of reference forms to choose (names, kinterms etc.) are governed by conversational preferences that shape “referential design”. Six preferences – a preference for associating the referent to the co-present conversationalists, a preference for avoiding personal names, a preference for using recognitionals, a preference for being succinct, and a pair of opposed preferences relating to referential specificity – guide speakers towards choosing a name on one occasion, a kinterm on the next occasion and verbal cross-reference on yet another occasion. Different classes of expressions better satisfy particular conversational preferences. There is a systematicity to the referential choices that speakers make. The interactional objectives of interlocutors are enacted through the regular placement of particular forms in particular sequential environments. These objectives are then revealed through the turn-by-turn unfolding of conversational interaction.

Doctor of Philosophy / Successful communication hinges on keeping track of who and what we are talking about. For this reason, person reference sits at the heart of the social sciences. Referring to persons is an interactional process where information is transferred from current speakers to the recipients of their talk. This dissertation concerns itself with the work that is achieved through this transfer of information. The interactional approach adopted is one that combines the “micro” of conversation analysis with the “macro” of genealogically grounded anthropological linguistics. Murriny Patha, a non-Pama-Nyungan language spoken in the north of Australia, is a highly complex polysynthetic language with kinship categories that are grammaticalized as verbal inflections. For referring to persons, as well as names, nicknames, kinterms, minimal descriptions and free pronouns, Murriny Patha speakers make extensive use of pronominal reference markers embedded within polysynthetic verbs. Murriny Patha does not have a formal “mother-in-law” register. There are however numerous taboos on naming kin in avoidance relationships, and on naming and their namesakes. Similarly, there are also taboos on naming the deceased and on naming their namesakes. As a result, for every speaker there is a multitude of people whose names should be avoided. At any one time, speakers of the language have a range of referential options. Speakers’ decisions about which category of reference forms to choose (names, kinterms etc.) are governed by conversational preferences that shape “referential design”. Six preferences – a preference for associating the referent to the co-present conversationalists, a preference for avoiding personal names, a preference for using recognitionals, a preference for being succinct, and a pair of opposed preferences relating to referential specificity – guide speakers towards choosing a name on one occasion, a kinterm on the next occasion and verbal cross-reference on yet another occasion. Different classes of expressions better satisfy particular conversational preferences. There is a systematicity to the referential choices that speakers make. The interactional objectives of interlocutors are enacted through the regular placement of particular forms in particular sequential environments. These objectives are then revealed through the turn-by-turn unfolding of conversational interaction.

This paper contributes to the discussion on the reference to ‘the people’ in contemporary politics. Illuminating the tension raised by representation, this corpus-assisted analysis investigates how Members of Parliament discursively construct the necessity to represent and involve the people while being at the same time the ones holding the floor. Close analysis of British, German, and French parliamentary debates on Europe reveals that parliamentary talk relies on a twofold dynamic: enacting the people’s voice and making parliamentary debates accessible. These patterns are represented in the speeches of all parliamentary groups. This shows that none of these discourse strategies are, per se, a prerogative of populist movements, thus shedding light on the necessity to think the articulation between populist style and populist ideology.

This study investigates, theoretically and empirically, the role of animacy in the development of gender systems. The theoretical background is a grammaticalisation approach to language change. Concerning gender, this presupposes that classifications begin as semantic distinctions in the realm of animacy with flexible, contextually based agreement between the gender-marking elements. This kind of gender is called contextual gender. In the course of time, these classifications will spread into other areas, they become desemanticized and the agreement relation grammaticalizes into one of government where the inherent gender of the head noun controls the gender of the agreeing elements, irrespective of contextual factors When this leads to a great number of violations of the principles of contextual agreement in the realm of animacy, a new cycle of semantic classification will begin, creating layers of classifications. For German and Swedish two different layers are discerned respectively. The empirical starting point of this project was the observation of two opposite developments in the area of female person reference in Germany and Sweden. As a consequence of feminist critique of language, mainly targeted at the use of socalled masculine generics, in Germany the use of female gender-specific nouns increased substantially, the major means being female derivation with –in, so-called motion. Although similar means for female derivation exist in Swedish, i.e. -inna and -ska, the number of derivations used is decreasing. In order to isolate socio-cultural and historical facts from language-internal mechanisms behind the diverging tendencies, a historical sketch of the development of equal rights, of language criticism and of the development of the female suffixes is drawn for the respective countries. It is obvious, that the German strategy to achieve gender-fair language use is established by making women visible by means of motion, while in Sweden the use of gender-neutral forms for a long period of time has been regarded as a sign of equality. This ‘neutral’ use of former masculine and male-specific forms has been made possible by the merging of the two nominal genders masculine and feminine into uter (Sw. utrum). A contrastive study of comparable German and Swedish newspaper texts shows that the lack of motion in Swedish is partly compensated by composition and attribution with gender-specific lexemes. Still, the 64% gender-specific noun phrases in Swedish cannot compare with the 95% in German. But the use of gender-specific forms for well over half of the person references calls into doubt the general opinion shared by most Swedes that Swedish has a gender-neutral person reference system. Linguistic asymmetry persists as long as gender-specification is restricted to one half of the gendered population, whatever the means for specification. The almost exclusive use of gender-specific forms in German is seen as indicative of a grammaticalisation process. Haspelmaths invisible hand explanation of grammaticalisation is used to show how the development of -in in German fulfils just about every requirement on a grammaticalisation process – language-external as well as -internal – while -inna and -ska neither are promoted sufficiently by the speech community nor does there exist a paradigm that could accommodate them. In contrast to Swedish, where the suffixes remain strictly derivational, it is demonstrated that -in is turning into an inflectional marker. The German gender sub-system for person reference is developing into a semantically based system with genderflexible person denominations. A study of the pronouns agreeing with non-personal-agents in a parallel corpus of EU-documents shows that other aspects than purely referential or formal ones impinge on the choice of agreement forms. Non-personal-agents in certain contexts expose both agency and intentionality, which turns them into suitable agreement partners for animate pronouns. In Swedish, all animate pronouns are sexed, leaving a “Leerstelle” for these inanimate but agentive and intentional referents. In German, this problem is covered by the polysemy of the personal pronouns. Non-personal-agents are shown to be one possiblesource for the spreading of a linguistic innovation from the realm of animacy into inanimate contexts via semantic and thematic roles that share important features with animates proper. The last study makes use of different types of German monolingual corpora in order to investigate the agreement between inanimate nouns with female inherent gender – from non-personal-agents and abstracts to concrete nouns – and agent nouns which can potentially expose agreement by female derivation. Although the results are rather heterogeneous, they allow the formulation of the hypothesis that agreement is more likely to occur with nouns for which a metaphorical bridge to stereotypical conceptions of femininity can be constructed and that key collocations with high frequency such as die Kirche als Trägerin or die DNA als Trägerin der Erbinformation contribute significantly to the spread of the agreement pattern.