The goal of my main research line (projects funded by NWO–VENI [€247.000]; Dutch Rehabilitation Fund [€50.000]) is to identify and chart the mechanisms behind the different forms of developmental speech sound disorders (SSD), and in particular subtype childhood apraxia of speech (CAS). More specifically, how do the phonological and sensorimotor impairments relate to core deficit, derived or consequential deficits, adaptive strategies, and how do they express themselves in symptomatology?

The first study addressed the interaction between speech motor control, phonology, and auditory self-monitoring in SSD. A first set of experiments investigated the ability to compensate and adapt for perturbed auditory feedback in children with SSD compared to age-matched normally developing children. The results showed that at the group level, the normally developing children were better able to compensate and adapt, adjusting their formant frequencies in the direction opposite to the perturbation, while the group of children with SSD followed (amplifying) the perturbation (Terband, van Brenk, & van Doornik-van der Zee, 2014). These results indicate that while most children with SSD were unable to compensate for perturbed auditory feedback, they could detect incongruences in auditory feedback and could adapt their target representations. A detailed look at the individual results revealed large individual differences lying underneath the group-results. Strong positive correlations were found between the amount of compensation and performance on oral motor movement non-word repetition tasks, tasks that are especially taxing on the (auditory-)motor system.

The finding of the children with SSD amplifying the perturbation was rather unexpected. A detailed analysis of the correlational results further revealed that especially the children with CAS were the ones that followed (amplifying) the perturbation. As to what would cause these children to follow (amplifying) the perturbation, my hypothesis is that due to weak phonological representations the perceived formant shift is interpreted by the speech production system as a target shift. This corroborates with data showing greater deterioration of speech output in speech apraxia when auditory feedback is masked, both in adults (Maas, Mailend, & Guenther, 2015) and children (Iuzzini, Hogan, Guarino, & Green, 2015), indicating a greater reliance on auditory feedback as was predicted by earlier modelling studies (Terband & Maassen, 2010; Terband, Maassen, Guenther, & Brumberg, 2009). In a follow-up computational modelling experiment, we managed to replicate the results of the behavioural auditory feedback perturbation experiments when weak forward control was simulated in the model (Terband, Rodd, & Maas, 2015, 2019). Together, this series of behavioural and modelling studies provide converging evidence that speech apraxia can be conceptualized as a deficit in feedforward control, with a greater reliance on feedback control as a derived deficit.

However, whether weak phonological representations are cause or consequence of weak feedforward control is not entirely clear. Previous behavioural studies have shown that phonological symptoms do not necessarily reflect phonological impairment but can also follow from acoustic-perceptual and/or articulatory-motor impairment. We investigated the effect of these factors on the build-up of phonological representations in a modelling study. The results of the computer simulations showed that articulatory-motor impairment lead to deviant phonological representations when combined with an intact ability of monitoring and correcting auditory errors, whereas it leads to a more general problem in auditory-motor integration when combined with acoustic-perceptual impairment (Terband, Maassen, Guenther, & Brumberg, 2014).

The hypothesis of participants following (amplifying) the perturbation due to weak phonological representations was further investigated in a series of follow-up studies in which we ultimately combined the feedback perturbation paradigm with detailed articulatory and perceptual measures that are indicative for phonological representations. This series of studies focuses on individual differences in a correlational design and features only participants with typical speech development, to separate the role of phonological representations from the role of derived deficits that are concomitant in disordered populations, and therefore focused on adults (which also enabled larger sample sizes). Results of a first experiment comparing children and adults indicated similar mechanisms and similar within-group variation (Terband & van Brenk, 2015; van Brenk & Terband, 2020). Data collection of the second experiment is completed and the data is currently being analysed.

Largely based on the experimental results and insights gathered in this project, we have formulated a model of speech development and disorders that can serve as a framework for a process-oriented approach to diagnosis and treatment planning of childhood speech disorders (Terband & Maassen, 2012; Terband, Maassen, & Maas, 2016a, 2016b). The framework leaves the current diagnostic classifications based on a description of behavioural symptoms and comprises two general diagnostic categories labelled “speech delay” and “developmental speech disorder” (Terband et al., 2016a, 2016b; Terband, Maassen, & Maas, 2019). Within these categories, treatment goals are formulated on the level of processes. A diagnosis comprises the assessment of two aspects; 1. characterisation of the developmental trajectory, and 2. identification of the processes involved and isolating the underlying core deficit. This process-oriented approach holds important advantages over the traditional method of diagnostic classification. It offers direct leads for treatment aimed at the specific underlying impairment tailored to the specific needs of the individual that is evaluated and adjusted in the course of the speech disorder.

The development of the Computer Articulation Instrument (CAI; Maassen et al., 2019) and its clinical validation (Diepeveen, van Haaften, Terband, de Swart, & Maassen, 2019; van Haaften et al., 2019) exemplifies a first important first step toward a user-friendly computer-based process-oriented assessment battery, yielding a comprehensive standardized and norm referenced speech profile that allows the assessment and characterisation of the developmental trajectory. The next step is to identify the processes involved and isolate the underlying core deficit(s) by means of a structured comparison of tasks and conditions to directly manipulate speech processes. The required experimental research methodology, e.g. masking noise, auditory perturbation, distorting kinaesthetic feedback, and short learning tasks (brief diagnostic therapies), has become available only recently, and we are currently working on their further development and refinement, and subsequently implementation.