Final Report Summary - MORLAS (Morphosyntactic language skills in deaf children with a cochlear implant: a cross-linguistic study on Dutch and German)
Marie-Curie Intra-European Fellowship
Morphosyntactic language skills in deaf children with a cochlear implant: a crosslinguistic study on Dutch and German (MORLAS) – Final Report
Nowadays, many profoundly deaf children are given access to auditory information by means of a cochlear implant (CI). Thanks to this device, these children are able to develop oral speech and language. However, there are still many open questions about the nature of speech and language skills of CI children. Previous research has focused mainly on perceptual achievements, while far less attention has been paid to speech and language production, i.e. to how well do CI children speak? Moreover, little is known about the “long term” development, i.e. what is the language achievement of CI children when they are at preschool age? This is especially relevant in view of the fact that more and more CI children are being integrated into the mainstream school system.
The aim of the MORLAS project was to investigate speech and language skills of CI children aged 4 to 6 years (preschool age). The project focused on CI children’s achievements in a major aspect of language, its morphosyntax. Gaining command of the full range of morphosyntactic features of a language requires highly sophisticated processing abilities and still constitutes a major challenge even for normally hearing preschoolers. Furthermore, the project compared CI children’s speech and language skills in two different languages, Dutch and German. This cross-linguistic perspective allowed to see whether CI children’s achievements vary depending on the specific language the child is exposed to.
The project progressed through three phases: In the first phase, the fellow worked on theory development of the relationship between subject-external linguistic factors and morphosyntactic development, as well as on the operationalization of linguistic factors that are relevant for morphosyntactic development in CI children. Focus was laid on the operationalization of factors of perceptual salience and distributional properties of noun plurals in Dutch and German. For this purpose, a novel framework highlighting similarities and differences in stem and suffix structures between the two languages (viz the method of “suffix predictability and stem transparency”) was used and further developed (Gillis et al., under review). The advantage of this method is that morphological operations in languages are being analyzed in terms of quantifiable notions, which makes cross-linguistic comparisons quantitatively feasible. At the same time, the fellow studied the experimental protocols used for the Dutch data collection in the project “Lexical and morphosyntactic development in young children with a cochlear implant” (Research Foundation – Flanders, FWO; project leader Prof. S. Gillis) and developed a German equivalent to the Dutch language assessment procedures.
In the second phase, the fellow collected and transcribed the German data of the project. The German study sample consisted of 10 children with a cochlear implant (CI) and an age-matched control group of 30 normally hearing (NH) children. A major challenge was the recruitment of a sample of German-speaking CI children who strictly met the selection criteria of the Dutch study sample: (a) diagnosis of profound hearing impairment in the first year of life, (b) cochlear implantation before the age of 24 months, (c) raised in normally hearing, monolingual families, (d) oral education, with or without the support of basic signs. Exclusion criteria were patent neurological disorders and/or general cognitive developmental delays. For this purpose the fellow established a new research collaboration with the ENT group at Vienna General Hospital (AKH) where testing of the German-speaking CI children took place, after successful application to the Ethics Committee of the Medical University of Vienna (EK.NR.1027/2013). The project employed a multi-task approach with three different language assessment procedures that vary with respect to the investigator’s control on the child’s linguistic output: i) experiment (controlled), ii) scripts (semi-controlled), and iii) conversation (free), with a total test duration of 40 minutes per child. In addition, parents were asked to fill out a short questionnaire on the child’s development.
In the third phase, the fellow coded and analyzed the German and Dutch data of the project, with special focus on noun plurals (viz the impact of “suffix predictability and stem transparency”). Statistical mixed-effect modeling of the experimental data showed only a slight crosslinguistic difference in success scores (p<.072), with Dutch-speaking children performing slightly better than German-speaking children. There were no significant differences in correct plural responses between CI and NH (p=.508). Importantly, both hearing groups showed the same patterning: higher success scores for plurals with a highly predictable suffix without a stem change, and the lowest scores for plurals requiring an exceptional suffix with substantial stem change (p<.001). However, there were significant differences in repetition of singulars (p<.001). These results suggest that at preschool age, early implanted children show even age-appropriate patterns of noun plural formation, but still have to catch up with respect to the ability to associate a given singular with its plural form. Results of the scripts and conversation data are currently being analyzed.
The training activities of the fellow focused on advanced statistical tools for data analysis. The fellow attended three statistics courses (Basic principles in statistics, R workshop, Multilevel Modeling) at the “core facility” StatUA, specifically dedicated at training researchers in advanced methods in statistics. Moreover, the fellow received intense face-to-face tutoring with respect to the advanced statistical technique of mixed-effects modeling with the statistical software R.