Introduction. The diversity and the scope of the responses to my target article (Kissine 2021) demonstrate the great potential linguistic research has for the field of autism. These responses also reveal, albeit sometimes implicitly, that a chief challenge in researching language in autism lies in the complexity and the subtlety of the data. In organizing my own response under this angle, I hope to react in a constructive way to most of the criticisms and suggestions voiced in these commentaries. In the next section, I return to the relationship between socio-communicative skills and language trajectories in autism; in particular, I argue that it is crucial to distinguish between predictors of language delays and those of language outcomes. Next, I discuss the complexity of pragmatics in autism, warning against reductionist attempts to subsume all data on pragmatics in autism under a single processing model. In the third section, I turn to the more general issue of the variability of individual profiles on the autism spectrum and ask how it can be integrated in a meaningful way within research on language in autism. I conclude this brief response to commentators by issuing a plea to think of the diversity of linguistic theories and schools as an opportunity for better understanding language in autism.
Joint attention and language trajectories.There is little doubt that sociocommunicative skills-including but not limited to joint attention-increase the opportunity for linguistic experiences and enhance the child's access to language. When such skills are significantly below the typical range, as is the case for many young autistic children, language acquisition is very likely to be delayed. For instance, as pointed out by Arunachalam, Artis, and Luyster (2021) in their response, a robust association between response to joint attention and language levels in autistic children emerges from the meta-analysis by Bottema-Beutel (2016). 1 However, the close association between atypically low joint attention and the likelihood for an autistic child to be non-or minimally verbal (see e.g. the studies on minimally verbal autistic children by Wodka et al. 2013or Yoder et al. 2015 does not entail that joint attention also predicts how language levels will evolve.Not carefully distinguishing between predictors of language delays and those of language outcomes may lead to significant mischaracterizations of language trajectories in autism. Take, as a first example, the longitudinal study by Anderson et al. (2007), which, according to the response by Goldberg and Abbot-Smith (2021:e172), shows that 'joint attention is predictive of autistic language development' (emphasis theirs). Anderson et al. followed three groups from the ages of two to nine years: autistic children, children with a diagnosis of pervasive developmental disorders not otherwise specified, and children with a nonspectrum developmental condition. At the study onset and independent of the diagnostic group, joint-attention scores correlated (along