Patterns in clitic pronouns: assessment of clitics in Italian in typical and non-typical populations

Clitic pronouns are linguistic elements which stand at the crossroads between different areas of the language faculty. They have specific morphology, syntax, and discourse functions. Use of this element requires the management of several aspects that draw from these areas of language. Clitics have been shown to appear early in typical acquisition, but to require longer for all aspects to be correctly managed, and they proved effortful in other modes of acquisition and atypical language, where they are clinical markers of impairment. In this work, we implement a comprehensive assessment of clitics aiming to investigate in what way different occurrences of this structure highlight different patterns of linguistic performance. To do so, we focus both on accuracy across conditions and on answer strategies, with an experimental protocol testing the following: comprehension of reference in binding constructions, production of clitics in two argument positions (direct object and indirect object), production of clitics in two sentence positions (preverbal and postverbal), production of clitics with different person features (1st/2nd and 3rd), and pragmatic abilities in the alternation with the lexical noun phrase and in perspective shift contexts. The assessment was tested on the following groups of native speakers of Italian: a group of pre-school and primary school typically developing children (Study 1), a group of adult heritage speakers of Italian living in the UK (Study 2), and four (pre)adolescents with a diagnosis of ASD (Study 3). Results found in this work showed that a comprehensive assessment of clitics can highlight similarities as well as differences in linguistic profiles according to different groups. Specifically, comprehension was not a discriminating factor in the populations: all populations tested showed to have access to abstract representations of clitics in binding constructions, as comprehension of simple clitics was generally unproblematic. An exception was found in the pre-schoolers, but this may have been due to task-related factors. In line with previous results on typical, atypical, and bilingual populations, production of Italian 3rd person direct object clitics with finite verbs showed different patterns across populations: in our data, typical children of all ages and ASD pre-adolescents showed to correctly produce this instance of the clitic the majority of the time, while heritage speakers of Italian showed poor production rates on this instance of the clitic. Importantly, our data shows similar results for indirect objects, showing that if the cliticization process is accessible, it is accessible regardless of the argument position occupied by the cliticised object. Another crucial result is that our studies find different patterns to be highlighted by the production of clitics on non-finite verbs as opposed to those on finite verbs: accuracy is similarly high in one high-performing ASD pre-adolescent and in the group results of 8-year-old children, but 4-year-old and 6-year-old children, as well as some ASD participants show chance or below chance performance in enclisis. The same was true for production of clitic combinations, although it was the least accurate structure across all groups, particularly in non-finite constructions. Here, heritage speakers and a few ASD speakers produce little to no instances of this construction, while those who produce it the most are the highest performing ASD participant and the oldest group of children. It is plausible to assume that the derivation of both enclitics and clitic combinations may require harder computation. These results allow us to conclude that if an individual or a population has issues on single DO and IO clitics with finite verbs, they will have issues with all other instances in the assessment. If an individual or a population shows no issues on single DO and IO clitics with finite verbs, they show typical language. In this, single clitics are coherent to their role of clinical markers. However, this study highlights the power of other instances of clitics, namely enclitics and clitic combinations, to unearth vulnerabilities to complex language. Types of non-target response also showed to be in part characterizing of different populations. The most striking result in this sense is the production of a lexical NP in place of the clitic. In typical development, this answer type only becomes the most used alternative answer in the hardest constructions, namely in enclisis; on the other hand, heritage speakers systematically use this construction as the alternative to clitics and use it more than they use clitics across all conditions. In our results, errors that are usually associated with impairment, particularly in younger participants, were marginally present in all groups. These are errors on the φ-features of the clitics, omissions, and misplacements (of which we found no instances), and they were limited to the youngest TD group, the heritage speakers, and the lowest-performing ASD participant. A recurrent alternative structure employed in conditions eliciting enclitics was the production of a simplified, finite verb structure.