Conversational alignment in children with an Autistic Spectrum Condition and typically developing children

Conversational deficits are often the first thing people notice about children with an autism spectrum condition (ASC), which affects around 1% of the population. Many ASC children experience communication difficulties and unrewarding interactions, and in turn people who interact with ASC children often find their conversations awkward. These conversational deficits have been linked to impaired theory of mind (ToM), the ability to attribute thoughts and feelings to others. But it is not clear exactly how these impairments might affect their communication, and particularly language production.

Our research investigates ASC children's conversational deficits by focusing on conversational alignment: the tendency for partners to imitate each other's use of language (e.g., word choice and grammar). Such alignment appears to be important for both effective communication and satisfying interactions. Our research examines whether ASC children show disturbed patterns of alignment, in ways that might explain some of their communication difficulties. We investigate whether ASC children spontaneously align with a conversational partner, and whether they do so in the same ways as typically developing (TD) children. By examining alignment under different conditions, we can draw inferences about the nature of their communicative impairments.

Our research focuses on the relationship between alignment and two factors where we might expect differences between ASC and TD children: 'audience design' and social-affective goals. Previous research suggests that although speakers may align just because they have heard a word or a structure before (and so have been 'primed' to re-use it), they also align in order to achieve particular goals, in ways that may implicate ToM. For example, they may adapt their language according to what they think their partner will best understand (audience design). They may also be guided by the desire to build a stronger relationship with their partner (i.e., social-affective goals), imitating their partner's language to express affiliation. ToM and social-affective impairments are common in ASC, and we hypothesise that ASC children may therefore showed a reduced tendency to align with a partner, compared to TD children. We consider this issue in two separate strands.

The first strand investigates whether and how far audience design influences ASC children's alignment. Specifically, we ask whether there is a relationship between alignment and ASC children's ability to adopt another person's perspective during language production. In TD children, perspective-taking is guided by inhibitory control (IC). ASC children have poor IC, which could explain why some who pass ToM tasks nonetheless display conversational deficits. We aim to discriminate effects of ToM and IC impairment on ASC children's communication.

The second strand builds on findings that typical speakers show a two-way relationship between affiliation and imitation: greater affiliation leads to more imitation and vice versa. We hypothesise that ASC children will show a weaker relationship between alignment (linguistic imitation) and social-affective factors, so that they will be less responsive to their partner's behaviour. This research will complement a limited literature on how affective impairments might contribute to ASC children's communication difficulties.

In sum, our research will deepen our understanding of whether and how different aspects of impaired social understanding impact ASC children's language processing in conversation. If ASC children align in an atypical way, this could help explain why and under what circumstances communication is difficult for them, and why their conversational partners in turn find their interactions odd and unrewarding. Our findings will also cast light on TD children's language production in conversation, and so are relevant to the study of alignment, and social imitation more broadly.

The power to communicate one's needs and views effectively is a fundamental requirement of social participation, one which is blocked for many individuals with an autism spectrum condition (ASC). The main potential for impact of this research lies in increasing understanding of the nature of communication difficulties, with the potential to inform educational and clinical interventions to support more effective communication and so to enhance quality of life for ASC individuals and relevant stakeholders in the community. The research findings will also have implications for understanding typical development of communication skills, and hence implications for policy and practice in early years teaching and provision for young children. There may also be implications for the design of automated dialogue systems which are widely used across the world.

Paediatricians, speech and language therapists, education professionals and policy makers: we will reach this audience via dissemination of findings through research briefings (e.g., to the UK Parliamentary Office of Science and Technology (POST), the Scottish Parliamentary Information Centre (SPICe), and the Scottish Cross Party Group on Learning Difficulties); magazine articles (e.g. TES Teaching Magazine, The Teacher), and talks at professional conferences with which we already have links, e.g. UK Literacy Association, British Educational Research Association, National Paediatric Neurodisability Network. The findings will be useful for interventions to support educational and clinical approaches to developing conversation skills in ASC children and adults.

Stakeholders in the ASC community: There are many stakeholders involved in concerns of individuals with autism: ASC individuals, their families and carers, health and education workers, teachers and support groups (e.g. Autism Sussex, National Autistic Society, Research Autism). These groups are often well-informed and engaged with research so it is important to communicate results in a timely and suitable manner.

Technology developers: Understanding of alignment is valuable to developers of automated chat agents, widely used in commerce and service industries: the research addresses sources and patterns of natural dialogue alignment that can inform development and improvement of automated agents.

Media: There is considerable media interest in autism, fostered by national, regional and local agencies such as the National Autistic Society, Research Autism, regional groups such as Autism Scotland and Autism Sussex, and local groups. Further, there are misconceptions about features of communication in autism, meaning that engagement with media about current approaches is both important and welcomed.

Academic beneficiaries: As outlined above, the immediate academic beneficiaries will be engaged through conference presentations and peer-reviewed publications, and hosting a symposium on alignment via the Experimental Psychology Society. In particular, alignment research is not well-integrated with related work in other highly-relevant areas, notably cognitive psychology, human-computer dialogue design, developmental psychopathology and the study of imitation in both social and developmental psychology: our proposal touches on each of these areas and will support links between these through targeted publication across an appropriate range of journals and conferences.