Learning to speak a second language is a challenging experience for the majority of language learners. "Your whole person is affected as you struggle to reach beyond the confines of your first language and into a new language, a new culture, a new way of thinking, feeling and acting" (H. D. Brown, 2000, p. 12). Thus, teaching to speak a foreign language is also a defying undertaking where, if we want students to use the language in class realistically and autonomously, speaking class activities need to be 'productive, purposeful, interactive, challenging, safe and authentic' (Thornbury 2007: 90). That said, there is a widespread assertion amongst language teachers that in class "students do not talk at all" (Bahrani & Soltani, 2012. p. 26). Within that context, this study investigates the factors affecting student engagement to speak (SEtS) in a language class, which is an unexplored concept - from the students' perspective - linked to the fields of Student Engagement (SE), Willingness to Communicate (WTC) and Communicative Language Teaching (CLT). A mixed methods approach was used to obtain the data, including a focus group, student and teacher interviews, participant observation insights, self reflections and a survey. The data was analysed by using Grounded Theory (GT) strategies (Charmaz 2014; Glaser 2007; Pace 2012) within an analytic autoethnography (Anderson 2006) to search for an emergent theory that could explain what engages students to speak in a language class. The study is framed within a constructivist approach to data collection and analysis, and the research includes the participation of 388 Australian National University undergraduate students of French, Spanish, German and Italian at different levels of instruction, 14 teachers and the researcher's own voice, supplemented by the relevant literature. The journey is paved with quotes of students' and teachers' words and with self reflective ethnographic analytical memos. The findings, obtained through the process of coding, categorisation and theoretical development of the qualitative data, complemented by the quantitative results of the survey, are grounded on students' lived experiences as well as the researcher's. They suggest that SEtS is a socio affective process underpinned by three interdependent dimensions involving the teacher's personal qualities, the course content and the classroom environment. The students' level of connectedness to those three dimensions affects their self confidence and either engages or disengages them to speak in class. This study shows that what really matters to students is still the human condition underpinned by affective and behavioural components such as teacher-student relationships, motivation and anxiety, and it serves as a stepping stone for further research in the area of language pedagogy and SEtS, and particularly on the role the personality of a teacher may play in second language teaching and learning.