Aims of this dissertation

In this chapter, the two main topics of this dissertation, i.e. speech intelligibility and identifiability in HI children were introduced. In the following chapters, five perceptual studies will investigate these two topics in primary school aged NH and HI children by means of hearing status categorisations, speech quality judgements and intelligibility measures while considering influencing listener factors such as their native language and the degree of experience with children’s speech.

The first part of this thesis (chapters 2 and 3) consists of two studies on speech intelligibility. Both studies aim at investigating whether the spontaneous speech of early implanted seven-year-old children with CI differs from that of their NH peers, yet with a different focus and a different type of measurement. Chapter 2 investigates the intelligibility of short spontaneous speech utterances of children with NH and CI through transcriptions. For this chapter, a new intelligibility measurement is applied, viz. entropy. Entropy calculates the degree of agreement in the transcriptions of the listeners and is especially suited for spontaneous speech since the measurement approach does not require a “correct model”

to calculate intelligibility. The entropy is calculated at the word level. A high degree of agreement indicates high intelligibility, whereas a low degree of agreement between the listeners indicates low intelligibility.

Chapter 3 elaborates on the spontaneous speech of children with NH and CI. This study assesses intelligibility by means of a visual analogue scale and focuses on two aspects. Firstly, the study aims at investigating the role of the assessed linguistic level by comparing the intelligibility of sentences

and that of longer stretches of speech. Secondly, the study examines the effect of the (calendar) year of implantation by comparing two groups of children with CI who are implanted ten years apart but whose chronological age and length of device use at the time of the recording is comparable.

The second part of this dissertation (chapters 4-6) consists of three studies on speech identifiability and aims to answer the question whether listeners can differentiate the speech of early rehabilitated HI children (CI and HA) from the speech of NH peers. Chapter 4 investigates whether listeners with a varying degree of experience with the speech of (HI) children (viz. audiologists, primary school teachers and inexperienced listeners) can differentiate between NH and HI children, i.e. children with a CI and children with an acoustic HA. For this purpose, a labelling task was set up in which listeners were required to label each utterance with what they thought the hearing status of the speaker of that particular utterance was. Chapters 5 and 6 apply a different approach towards speech identifiability. More specifically, in these studies, the research question is whether listeners perceive a difference in the overall speech quality of children with CI, HA and NH. In two comparative judgement tasks, listeners indicated for pairs of stimuli which stimulus sounded best, i.e. had the higher overall speech quality. For these two studies, the methodological approach is identical, yet the listener panels differed. In chapter 5, the participating listeners belonged to the same groups as in chapter 4, i.e. audiologists, primary school teachers and inexperienced listeners. Chapter 6, on the other hand, investigates whether listeners unfamiliar with Dutch judge the speech quality of HI children differently

from native listeners. Non-native listeners without knowledge of Dutch will only consider speech aspects and will not be influenced by, for example, regional characteristics. Therefore, this study includes the comparative judgements of native speakers of Italian, French and German and compares those to the native Dutch speakers’ judgements.

Chapter 7 provides an overview of the main findings and their relevance to the field of linguistics in general and specifically, the domain of hearing-impaired children’s speech. Moreover, the chapter formulates some suggestions for future research. Finally, chapter 8 presents an English and Dutch summary of this dissertation.

Chapter 2

2. Measuring spontaneous speech intelligibility using entropy: early cochlear implanted children versus their normally hearing peers at seven years of age

Note: A version of this chapter was submitted for publication as a research paper. Reference: Boonen, N., Kloots, H., Nurzia, P., & Gillis, S. (under review). Spontaneous speech intelligibility: early cochlear implanted children versus their normally hearing peers at seven years of age. Journal of Child Language.

Abstract

Studies on the spontaneous speech intelligibility of children with a cochlear implant (CI) predominantly use rating scales. However, early implanted children with several years of device experience have been shown to reach ceiling scores. Therefore, despite the lack of a model transcription, the present study investigated children’s spontaneous speech intelligibility using transcriptions. More precisely, short utterances from a picture cued storytelling task of sixteen seven-year-old children with CI and sixteen normally hearing (NH) peers were transcribed by 105 listeners. Subsequently, altogether 6,720 transcriptions were analysed in terms of the degree of uniformity (i.e. entropy) between the transcriptions.

The results show that listeners agree more strongly on the transcriptions of NH children’s utterances than on those of children with CI, indicating

that the latter group has a lower intelligibility. However, the majority of CI children reach scores within the range of NH children’s scores. Moreover, older children reach higher intelligibility scores.

Keywords: Intelligibility; Spontaneous speech; Children with a cochlear implant