The role of misidentification of own speech in developmental stuttering

RESEARCH QUESTIONS: How do children learn to perceive speech? Does misidentifying their own speech cause children to stutter? METHODS: Approach 1: We will conduct a longitudinal study of children who stutter. Children typically begin to stutter between ages of 3 to 4, and most will recover by the age of 7 (Yairi & Ambrose, 2013). This enables a repeated measures comparison between children who stutter (CWS) and children who recover. We will measure: Auditory Brainstem Response to speech (sABR). This is an account of electrophysiological activity in neural pathways closest to the ear. Vestibularevoked myogenic potential (VEMP). This is an indirect measurement of activity in the vestibular system. We will work with two cohorts. Cohort 1 will be recruited from local speech and language therapy (SLT) services, with matched controls recruited from the ESRCfunded LuCiD project. We will follow this cohort from the age of 3to4 at which stuttering is typically referred to SLT services. Cohort 2 will be recruited from the LuCiD Language 0-5 study. This will enable us to investigate reports of stuttering from as early as 2-3, which often are not referred to SLT services (Yairi & Ambrose, 2013). Approach 2: With adults who stutter (AWS) we will manipulate auditory feedback during speech. Alterations to auditory feedback (AAF) have long been found to reduce severity of stuttering (Bloodstein, 1995). In conjunction with the CASE partner (Interacoustics), we will develop custom circuitry enabling AAF manipulations not previously tested. We will also measure sABR and VEMP as per approach 1. RATIONALE:
The 80% recovery rate from childhood stuttering (Yairi, 2013) is not fully understood. Many studies indicate abnormal auditory function in stuttering (Bloodstein, 1995). We hypothesise that misidentification of selfgenerated vocalisation (SGV) contributes to stuttering. There is no widely accepted mechanism for identification of SGV. However, the vestibular system, a part of the ear important for balance, can detect sound. The vestibular apparatus is too insensitive 3 to detect most environmental sounds, but is close to the larynx and may thus be able to detect the high sound pressures in the larynx during SGV. This would be via conduction through the body (BC), rather than through air (AC), and is at present untested. The ear normally detects sound via the cochlea. Thus, a study of SGV identification must evaluate three ascending routes: air conduction via the cochlea (ACC), and body conduction via both the cochlea and vestibular system (BCC/BCV). We hypothesise that the vestibular system contributes to identification of SGV, and more precisely that SGV is identified from synchronous ACC, BCC and BCV auditory stimuli. Our VEMP and s-ABR measurements will enable us to appraise this. Between ages 5-8 the auditory brainstem is still developing, and the s-ABR shows a temporary overshoot in latency and amplitude (Skoe et al, 2015). This overlaps with the age (6-7) by which most children recover from stuttering. There is no study of s---ABR in CWS, but two studies of s-ABR in AWS give identical results of increased latencies in waves V, A and O (Ibraheem & Quriba, 2014; Tahaei et al, 2014). We predict this pattern will be visible in the children who persist in stuttering, but absent in children who recover and fluent controls. The pattern indicates timing abnormality in onset and offset responses. Originality: According to the hypothesis of Howell and Sackin (2002), fluency in AWS will be most enhanced when additional synchronous stimuli are present. A vestibular response to SGV would be highly synchronous. This prompts four further hypotheses: H1. In normal speakers, SGV is identified from synchronous air--- and bone---conducted auditory streams in the cochlea and vestibular system. H2. Misidentification of self---generated vocalisation (SGV) contributes to stuttering. H3. The 80% recovery rate from childhood stuttering corresponds to auditory brai