Résumé : Females have been historically under-represented in autism research, leading to a male-centered understanding of autism and contributing to a sex disparity in prevalence. Growing evidence suggests that this gap partly stems from the misidentification of autistic females. However, a key paradox is the lack of data on undiagnosed females. Little is known about the characteristics of late-diagnosed females before adulthood and studying only diagnosed females may obscure the very characteristics that lead others to go unnoticed. Among factors that could explain delayed diagnosis, previous studies have identified a pragmatic advantage for autistic females. This dissertation explores potential sex differences in pragmatic abilities through two complementary approaches: (1) comparing diagnosed autistic females with autistic males and non-autistic individuals, and (2) examining sisters of autistic individuals as a population at increased likelihood of autism who may thus be diagnosed later. Chapter 1 examines sex differences in narrative coherence, analyzing story grammar, connectives, references to characters and internal state language. Autistic individuals produce less coherent narratives than non-autistic individuals, with no sex differences. Chapter 2 investigates silent and filled pauses in non-interactive and interactive speech. Autistic participants rely more on silent pauses than non-autistic participants, and females produce more pauses overall. Chapter 3 extends this analysis to discourse markers and finds no group nor sex differences, although autistic males produced fewer distinct discourse markers than non-autistic males. Chapter 4 focuses on sisters of autistic individuals: narrative coherence, pause production, and caregiver-reported pragmatic difficulties are analyzed. Caregivers report greater difficulties in pragmatics for sisters compared to non-autistic individuals, and some sisters face challenges in the narrative task.Overall, this dissertation contributes to efforts to better characterize autism in females. Our results do not support a pragmatic advantage for autistic females, possibly because autistic females in this dissertation were diagnosed early and resemble autistic males. However, Chapter 4 highlights the value of including in autism research females at increased likelihood of autism: expanding research populations may help capture the variability of autism in females and improve identification strategies