"Speech and hearing disorders"
June 10, 2005, Max Planck Institute (MPI), Nijmegen, http://www.mpi.nl
Organized by the Dutch Society for Phonetic
Pete Howell, Psychology Department, University College London
Levelt's (1989) perceptual loop theory continues to have an enormous impact on theorizing about speech control covering, through the work of Kolk and Hartsuiker, aphasia as well as stuttering. The Levelt theory links perception and production, maintains errors in production are detected by the perceptual system which interrupts and restarts speech (resulting in certain forms of disfluency). This talk starts by giving grounds for disregarding the view that the perceptual system is involved in production control. An alternative account of speech control and its failure is offered where planning and execution processes are independent and problems in fluency control arise when they are not in synchrony (in particular, when the speaker wants to start execution before the complete plan is available). The theory predicts that fluency will break down when speech rate occurs on difficult utterances. In this contribution, we offer newly developed techniques for measuring speech rate and word difficulty and evidence about how these interact. We also show ANNs can learn where fluency is likely to fail using these measures. Finally we offer evidence from FMRIB scans on speakers who stutter that support increased activation in the structures deemed to be responsible for rate control, and also evidence in favor of the theory from a priming task.
Patricia Gulpen, Ben Maassen, Lian Nijland Paediatric Neurology, Medical Psychology; Radboud University Nijmegen Medical Center
Phonetic transcriptions of pathological speech are used in research, but it is not common practice to make phonetic transcriptions for the clinical diagnosis of developmental speech disorders. In the past, this had led to a number of misconceptions, such as the idea that children with developmental apraxia of speech (DAS) produce many anticipations and perseverations (see Thoonen et al. (1994) and Thoonen (1998) for counterevidence). In a series of clinical studies, departments of the University Medical Center in collaboration with the Dept. of Language and Speech, Radboud University Nijmegen, and the HAN University, Dept. of Speech and Language Therapy, have conducted comparative studies of different speech disorders in children using phonetic transcriptions. The first series of studies concerned the within and between transcriber agreement, and the validity of phonetic transcriptions (Vieregge & Maassen, 1999). It was shown that only rather broad phonetic transcriptions, with highly restricted use of diacritics, reaches sufficient agreement for clinical purposes. Further, recommendations to increase agreement could be derived from the different transcription methods used. The second series of studies concerned the differential diagnostic value of phonetic transcriptions. Different error patterns were found between dysarthrias on the one hand, and phonological disorders and DAS on the other. Differentiating between phonological and motor planning disorders seems to require additional acoustic and/or kinematic analyses (Nijland, 2003).
Protocols for transcribing and analyzing speech samples are implemented in LIPP (Oller, 1998); examples will be demonstrated.
Esther Janse, Utrecht institute of Linguistics, OTS
Research has shown that Broca's and Wernicke's aphasic patients show different impairments in auditory lexical processing. To investigate these impairments further, two word recognition experiments were set up with two groups of aphasic listeners and one age-matched healthy control group. One experiment was designed to investigate the process of lexical competition; the other experiment was aimed at lexical activation of word candidates.
A speeded lexical decision experiment with form-overlapping primes (word-initial overlap) was set up to investigate lexical competition. The results of this experiment showed inhibition/competition effects for control adults and Broca's aphasic patients, but facilitation effects were found for Wernicke's aphasic patients. This indicates that Wernicke's aphasic patients are mainly impaired in suppression of once-activated word candidates and selection of one winning candidate, which may explain their problems in auditory language comprehension.
Broca's aphasic patients, on the other hand, are claimed to have problems in the activation of lexical candidates from acoustic-phonetic input. A second speeded lexical decision experiment was set up to investigate the influence of neighbourhood density on response times to nonwords for the two patient groups, compared to the healthy control group. Two groups of nonwords were presented to find out whether longer response times (and lower correct scores) would be found for nonwords with a large number of neighbours than for nonwords with few neighbours in the mental lexicon. Neighbourhood density effects were found for all three listener groups. The healthy control and Broca's aphasic patient group showed density effects on mean response time and correct scores, but the Wernicke's aphasic patients only showed a density effect on the correct scores. Implications of these findings will be discussed.
Irma Verdonck, VU University Medical Center, Department of Otolaryngology
In the Netherlands approximately 2500 patients are diagnosed with primary head and neck cancer (HNC) every year. Compared to other cancer patients, HNC patients have to deal with severe side effects of treatment: in addition to general complaints as pain and fatigue, fear for tumour recurrence and psychological distress, patients are often confronted with facial changes, and voice, speech, or swallowing problems. Next to oncological outcome (tumour control, survival rate), quality of life and functional outcome have become more and more important. The past 10 years have shown an improvement in the treatment possibilities regarding functional outcome. For example, free flaps such as the radial forearm flap have become the preferred method of surgical reconstruction for larger defects in the oral cavity and oropharynx because of reliable reconstruction and because of improved functional characteristics of dynamic structures such as the tongue and pharynx. Another example is the development of a sound-producing voice prosthesis for laryngectomized patients. In 1999 a national CBO guideline was developed by the Dutch Working Group of Head and Neck tumours on diagnosis and treatment of laryngeal cancer followed by a guideline on oral and oropharyngeal cancer in 2003. In both guidelines the importance of functional (voice and speech) outcome is stressed. Unfortunately, comparison of the literature on voice and speech outcome is limited due to a diversity of voice and speech analysis methods. This diversity is caused by the various viewpoints of which voice and speech can be approached. Also, there is an ongoing improvement of analysis methods.
At the department of Otolaryngology – Head & Neck Surgery of the VU University Medical Center (VUmc), several projects on HNC cancer patients have been carried out or are ongoing. The leitmotiv in the various projects is functional (voice, speech, swallowing) outcome in relation to health related quality of life. The main aim is to improve diagnosis, treatment and rehabilitation options. The associated aim is to improve voice, speech, and swallowing assessment methods. For this purpose a “Speech Lab” has been developed, which is now in use in clinical routine as well as in research projects. In the presentation, an overview will be given on a few selected studies including voice and speech outcome in relation to psychological distress and quality of life of HNC patients.
Petra Jongmans, Frans Hilgers,
Louis Pols, & Corina van As
In cases of laryngeal cancer, it is sometimes necessary to perform a total laryngectomy. This procedure changes the anatomy and physiology of the vocal tract, with the most noticeable effect on speech. By applying a voice prosthesis, enabling the patient to use tracheoesophageal (TE) speech, speech is of better quality than with esophageal or electrolarynx speech, but still very deviant from laryngeal speech. In different countries, studies investigating the intelligibility of TE speech have been performed. In the Netherlands, only voice quality has been studied in detail. For the understanding of the physiology of the neoglottis and for improving the results of speech therapy it is important to study intelligibility as well. In our pilot study, we employed different methods to investigate intelligibility.
We collected speech material of 11 laryngectomees. Ten naïve and four experienced listeners evaluated this material perceptually. This perceptual analysis shows us which speech sounds are difficult to produce and which confusions are often found. We have also performed acoustic analyses on some of the speech samples. We wish to link the results of the perceptual and acoustic analyses to the physiology and anatomy of the neoglottis and vocal tract and for that reason we have also made recordings of the neoglottis with a digital high speed camera and used videofluoroscopy. We are particularly interested in the voiced/voiceless contrast as the (in)ability to produce this contrast will tell us more about the working of the neoglottis and the amount of control TE speakers have over their voice source. In this talk, I will discuss the first results of the perceptual and acoustic analyses and I will show some examples of digital high speed imaging.
Ellen Gerrits, Maastricht University
This paper reports the design and set-up of a study on speech perception of young deaf children with a cochlear implant. A cochlear implant (CI) is an appropriate sensory aid for selected deaf children who receive minimal benefit from conventional amplification. Many children using cochlear implants have acquired speaking and listening skills and have developed a spoken language system which is beyond what previously could be achieved with hearing aids. However, there are large individual differences in language performance among deaf children with cochlear implants and their language skills continue to lag behind those of normal-hearing children. It is important to note that even today, it is not really understood what the key
to success is.
M.C. Langereis, Viataal, St. Michielsgestel, and
Profound deafness has a devastating effect on the development of spoken language and speech in children. Cochlear Implants (CI) provide children with auditory input necessary for the development of speech and language production. So far, relatively much research on cochlear implantation in children has been directed towards the effect on the perception of speech. Recently, research has suggested (Svirsky, 2003) that congenitally deaf children may be able to develop expressive and receptive language skills at a normal pace.
In this presentation we will present the auditory and language data for two groups (1-2;6 years and 2;7-4;0 years) of children at two and three years post-implantation. The results demonstrate that the group children implanted at a younger age develop speech perception and language skills at a higher pace than the older group. Furthermore, these younger children start with a smaller language gap pre-implantation. The results will be related to those in literature. In addition, some results concerning vowel production of children with cochlear implants will be presented.
Johanna Barry, Department of Experimental Psychology, University of Oxford
In this paper, I describe a study investigating the relationship between tone production and perception for two groups of Cantonese-speaking children: a group of implant users and a group of normally-hearing children. Tone production involved producing citation form tones as part of a picture-naming task. Responses were recorded and analysed acoustically. Results were summarized as ‘size of tonal space’. The tone identification task was a two alternative forced-choice task, the children listened to a word presented through a speaker and then pointed to the corresponding picture from a choice of two possible options. A positive correlation (p<0.001) was found between size of tonal area and mean number of tones correctly identified. Most implant users had reduced tonal spaces relative to the normally-hearing children and performed poorly at the tone identification task. A few implant users however identified tones as successfully as their normally-hearing peers. They also had tonal areas that matched those used by the normally-hearing children. Since the implant does not provide much useable information about tone to users, it suggests that acquisition of tone production skills has played an important role in helping these children develop reasonable tone perception abilities. These results suggest that to compensate for shortcomings in current cochlear implant technology, tone production training should be emphasised in speech habilitation programs.
Astrid van Wieringen1, Robert P. Carlyon2 , Olivier Macherey1, Jan Wouters1
Psychophysical measures will be presented of different types of symmetric and asymmetric charge-balanced waveforms in human implantees. Data of the Laura and Clarion devices show that thresholds are significantly lower for pseudomonophasic waveforms than for the biphasic waveforms for 50% of the tested subjects, while thresholds of monophasic stimuli (with anodic and cathodic pulses alternating), are significantly lower for all subjects. The results demonstrate that thresholds and dynamic ranges of human cochlear implant users can be varied considerably by manipulating the way in which the charge produced by the initial phase of an electrical pulse is recovered.