Homonegative discrimination such as the denial of leadership qualities and higher salaries concern not only lesbians and gay men but also individuals who were perceived as lesbian or gay (Fasoli et al., 2017). Hence, it is assumed that especially straight people become victims of homonegative discrimination (Plöderl, 2014). The perception of sexual orientation is indeed stereotype-driven (e.g., Cox et al., 2015) but there is a lack of knowledge on how accurate stereotypes are – particularly those referring to speech. Despite a variety of sociophonetic and social psychological research related to sexual orientation and gender, an encompassing understanding is missing on how sexual orientation is expressed and perceived. The present thesis aims to fill these gaps. The two major aims of the present work are a) the examination of the accuracy of speech stereotypes in the context of sexual orientation and b) the development of a model on how sexual orientation is interpersonally construed. Overall, the present thesis comprises five manuscripts with the following aspects in common: They integratively deal with social psychological and sociophonetic perspectives, share a social identity approach, and primarily center speech instead of facial appearance. Moreover, mostly German and German native speaking participants, respectively, have been investigated. Manuscript 1 establishes the Traditional Masculinity/Femininity-Scale as a reliable and valid instrument for assessing gender-role self-concept. The invention was necessary because existing scales insufficiently represented the self-ascribed masculinity/femininity yet (e.g., Abele, 2003; Evers & Sieverding, 2014). Manuscripts 2, 3, and 4 address the (in)accuracy of speech stereotypes regarding stereotypic content and suggested within-group homogeneity. This is carried out by the application of different methodological approaches. On the one hand, relevant acoustic parameters of lesbian/gay and straight women and men were averaged for each group. On the other hand, voice morphing was applied in order to create prototypical and naturally sounding voice averages (Kawahara et al., 2008). Lesbians and straight women differed in none, gay and straight men in one of the analyzed acoustic parameters only. In contrast, a fine-grained psychological analysis yielded various evidence for acoustic within-group heterogeneity. In particular, the exclusivity of sexual orientation and gender-role self-concept have been acoustically indexicalized which suggests that speech stereotypes are inaccurate. However, voice averages do carry perceivable sexual orientation information. Hence, speech stereotypes can be considered as exaggerations of tiny kernels of truth. In Manuscript 5, previous literature on the interpersonal construction of sexual orientation is integrated in a model: The Expression and Perception of Sexual Orientation Model (EPSOM). This model postulates an indirect route and describes how sexual orientation information is transmitted from producer to perceiver by proposing three mediating components. Thereby, the model is able to offer an explanation why sexual orientation can be perceived with above-chance but far-away-from-perfect accuracy. Overall, the present thesis provides meaningful impulses for enhancements of research on social markers of sexual orientation and gender. This thesis offers a model on how sexual orientation is expressed and perceived, shows the benefits of combining sociophonetic and social psychological approaches, and points out the value of applying novel methods and technologies. Beyond that, the present thesis offers useful implications for practice. Speech stereotypes in the context of sexual orientation can be rejected as inaccurate – for example, native German straight men do not nasalize more or less than gay men. Thereby, the present thesis contributes to an erosion of stereotypes and a potential reduction of homonegative discrimination.