Abstract
Purpose and background
Acoustic signals are transmitted through the external and middle ear mechanically to the cochlea where they are transduced into electrical impulse for further transmission via the auditory nerve. The auditory nerve encodes the acoustic sounds that are conveyed to the auditory brainstem. Multiple brainstem nuclei, the cochlea, the midbrain, the thalamus, and the cortex constitute the central auditory system. In clinical practice, auditory brainstem responses (ABRs) to simple stimuli such as click or tones are widely used. Recently, complex stimuli or complex auditory brain responses (cABRs), such as monosyllabic speech stimuli and music, are being used as a tool to study the brainstem processing of speech sounds. We have used the classic ‘click’ as well as, for the first time, the artificial successive complex stimuli ‘ba’, which constitutes the Greek word ‘baba’ corresponding to the English ‘daddy’.
Patients and methods
Twenty young adults institutionally diagnosed as dyslexic (10 subjects) or light dyslexic (10 subjects) comprised the diseased group. Twenty sex-, age-, education-, hearing sensitivity-, and IQ-matched normal subjects comprised the control group. Measurements included the absolute latencies of waves I through V, the interpeak latencies elicited by the classical acoustic click, the negative peak latencies of A and C waves, as well as the interpeak latencies of A–C elicited by the verbal stimulus ‘baba’ created on a digital speech synthesizer.
Results
The absolute peak latencies of waves I, III, and V in response to monoaural rarefaction clicks as well as the interpeak latencies I–III, III–V, and I–V in the dyslexic subjects, although increased in comparison with normal subjects, did not reach the level of a significant difference (p < 0.05). However, the absolute peak latencies of the negative wave C and the interpeak latencies of A–C elicited by verbal stimuli were found to be increased in the dyslexic group in comparison with the control group (p = 0.0004 and p = 0.045, respectively). In the subgroup consisting of 10 patients suffering from ‘other learning disabilities’ and who were characterized as with ‘light’ dyslexia according to dyslexia tests, no significant delays were found in peak latencies A and C and interpeak latencies A–C in comparison with the control group.
Conclusions
Acoustic representation of a speech sound and, in particular, the disyllabic word ‘baba’ was found to be abnormal, as low as the auditory brainstem. Because ABRs mature in early life, this can help to identify subjects with acoustically based learning problems and apply early intervention, rehabilitation, and treatment. Further studies and more experience with more patients and pathological conditions such as plasticity of the auditory system, cochlear implants, hearing aids, presbycusis, or acoustic neuropathy are necessary until this type of testing is ready for clinical application.
1
Introduction
Learning disabilities are related to understanding or in using language, spoken or written, which may manifest in an imperfect ability to listen, speak, read, write, spell, or do mathematical calculations. Dyslexia is a condition characterized by unexpected difficulty in learning to read, impaired discrimination of rapidly presented acoustic stimuli, impaired detection of rhyme, and mild deficits in speech perception but with normal intelligence . Auditory brainstem responses (ABRs) provide a sensitive non-invasive neurophysiological method for selectively ‘tapping’ an individual’s speech processing capacity, which, because of auditory, linguistic and/or cognitive reasons cannot be reliably assessed using standard available measures . Studies have also shown that ABRs can be a powerful tool in the study of cognitive development and its disturbances . The ABRs provide a recording of the electrical events that occur along the auditory pathway of the brainstem. ABR peaks I through V are most likely generated by the cochlear nerve, cochlear nucleus, superior olive, lateral lemniscus, and inferior colliculus, respectively . Thus, by analyzing ABRs, it is possible to determine abnormalities in the conduction of action potentials along the nerve axons or in the synaptic activity of the neural brainstem generators. In this context, ABRs have been used for screening for acoustic neuroma , evaluation of vestibular evoked myogenic potentials , screening for neonatal hearing loss , and screening for multiple sclerosis . Abnormal ABRs have been found also in ischemic heart disease , drug abuse , diabetes , trauma , hyperlipidemia , systemic lupus erythematosus , schizophrenia ( and dyslexia . Other studies in individuals with dyslexia have shown normal ABRs to click stimuli . In this particular early study among eight dyslexic adults, normal ABRs were found in six. Recently, ABRs have been used in children while under anesthesia, where significant delays were found , and in neurodiagnosis for determining the site of a hearing lesion .
Non-speech stimuli, which are most often used in ABRs, such as ‘clicks’ or ‘tones’, can trigger a synchronic response from a large number of neurons and have a broad frequency, and the response provides information about the brainstem nuclei along the ascending auditory pathway . Brainstem responses to such simple stimuli have been widely used in clinical practice in the evaluation of auditory pathway integrity . However, non-speech stimuli do not provide insight about the actual processing of speech sounds. For this reason, brainstem responses elicited by more complex stimuli or complex auditory brain responses, such as speech and music have been used experimentally . Since speech is a complex signal varying in many acoustic parameters over time, studying the encoding of speech sound at the brainstem is of paramount importance. A complete description of how the auditory system responds to speech can only be obtained by using speech stimuli.
So far, only monosyllabic speech stimuli, such as ‘da’ and ‘ma’ have been used in order to study ABRs. Monosyllabic speech stimuli have elicited abnormal ABRs in populations with speech perception deficits such as learning problems , and in children with phonological disorders . King et al. studied complex auditory brainstem responses (cABRs) for the speech stimulus ‘da’ in children with normal development and children with learning disabilities and found significant brainstem processing delays. We have found a number of significant differences in the onset response of speech stimulus ‘ma’ in young adults with dyslexia compared with their normal counterparts.
In the present study, we postulated that a specific disyllabic speech stimulus like ‘baba’, corresponding to the English word ‘daddy’, and used in ethnic spoken language would be more appropriate in eliciting cABRs in Greek-speaking subjects. Disyllabic stimuli corresponding to familiar words induce a pattern of voltage fluctuations in the brainstem, resulting in a familiar waveform and yielding better information about brainstem nuclei along the ascending central auditory pathway. Therefore, for our patients, we have used the classic ‘click’, as well as, for the first time, the artificial successive complex stimuli ‘baba’. The disyllabic word ‘baba’ contains frequencies closer to ‘da’ than ‘ma’ and has components acoustically similar to ‘da’ . Furthermore, the word ‘baba’ is also used in many ethnic languages, such as Chinese, Arabic, Persian, Slavic, and Turkish and in other languages spanning from West Africa to the Eastern Mediterranean and all the way to East Asia.
2
Material and methods
In the Greek language, the word ‘baba’ corresponds to the English word ‘daddy’ and consists of two repeated ‘ba’ sounds. Thus, we used verbal acoustic stimuli consisting of broad band syllables ‘ba-ba’ with a fast rise, an inter-syllable plateau, and fall time, replacing the well established ‘da’ for similar studies in English-speaking subjects . All stimuli were created on a digital speech synthesizer . The acoustic properties of the stimulus were checked with a sound analyzer before proceeding with the experiment. To confirm that the stimulus met the desired specifications, the synthetic sound was acoustically analyzed in Praat . The final recordings were analyzed using a signal processing tool developed with MATLAB 6.0 for Windows and the peaks were named in the same way as those of cABRs .
2.1
Patients
Twenty subjects institutionally diagnosed as dyslexics were studied. All subjects were Greek speakers with Greek as their first language without a history of brain damage, language problems, psychiatric symptoms, or visual problems. Eight were males and 12 were females. They were between 18 and 23 years of age, with a mean age of 20 years. The control group consisted of 20 age- (19–22 years, mean 20.5), sex-, education-, hearing sensitivity- and IQ-matched healthy young adults. The IQ was estimated in our Institutions, from four index scores (verbal comprehension, perceptual reasoning, working memory, and processing speed) from the Wechsler Adult Intelligence Scale (WAIS IV) . Subjects with an IQ below 85 were excluded.
All subjects underwent a thorough audiological examination. Audiometric tests included speech reception threshold testing, pure-tone audiometry, speech discrimination testing, and tympanometry.
2.2
Broad band ‘fast’ and complex ‘long’ stimuli
In all subjects, the ABRs were recorded with the person in the supine position in a quiet room. All tests were performed in all subjects using auditory stimuli of broad spectrum (click) with an intensity of 80 dB nHL, repetition rate 11.29/s and frequency filters 100–3000 Hz. The parameters evaluated were the latencies for the waves I to V, and the interpeak latencies I–III, III–V, and I–V .
Since broad spectrum stimuli evoke brainstem potentials by the synchronous firing of more individual neural units, these potentials are very clear . However a fast-onset stimulus contains transient acoustic energy only at various frequencies, with frequency specificity in dyslexia. In order to eliminate such transients and obtain a good stimulus specificity, the cABRs of the subjects were also tested, using ‘complex’ acoustic stimuli with sufficiently long onset time and wider frequency band .
2.3
ABR acquisition
A PC-based stimulus delivery system controlled the time of delivery, stimulus sequence, and stimulus intensity, and triggered the PC-based evoked potential averaging system. All electrode impedances were < 5 kΩ. A MATLAB for Windows-based software tool identified and marked the onset, offset, and peak latencies and a trained observer resolved discrepancies on the PC-output of the system signal. Reproducibility of the results was always achieved. The cABRs were collected according to the method described before, in response to a click (0.1 ms) and randomly presented alternating polarities of ‘ba’. Alternating polarities were added together to isolate the neural response from that of the cochlear microphonic. cABRs were differentially recorded from Cz-to-ipsilateral earlobe, with the forehead as ground. The sampling rate was 20,000 Hz. Responses were bandpass filtered on-line from 100 to 2000 Hz. Sweeps with noise levels that exceeded ± 30 μV were rejected from the average. Three repetitions of 1000 sweeps each were collected in response to the click as well as for each polarity of ‘ba’. The click stimuli were presented at 80 dB SPL with an interstimulus interval (ISI) of 32 ms; the recording window was 20 ms, including a 10 ms prestimulus period .
The ‘ba’ stimuli were presented at 80 dB SPL with an ISI of 51 ms; the recording window was 60 ms, including a 10 ms prestimulus period . The latencies of the click-evoked waves I–V and the negative peaks following the stimulus ‘ba’, marked A and C, were compared with normative values. Negative peak A is the onset of the response to the stimulus and peak C is the initial part of the frequency following the onset of the response ( Fig. 1 ). We used the latencies of waves A and C as well as the interpeak latencies A–C, and compared them between normal and dyslexic subjects. On this basis, a multiparametric analysis (filtering, peak identification, and latency calculation) through a PC-based signal processing tool implemented with a MATLAB tool of the recordings of the examined normal and dyslexic cases was performed. The possible significant differences between the two methods in both groups were assessed using Student’s t-test for unequal samples .
2
Material and methods
In the Greek language, the word ‘baba’ corresponds to the English word ‘daddy’ and consists of two repeated ‘ba’ sounds. Thus, we used verbal acoustic stimuli consisting of broad band syllables ‘ba-ba’ with a fast rise, an inter-syllable plateau, and fall time, replacing the well established ‘da’ for similar studies in English-speaking subjects . All stimuli were created on a digital speech synthesizer . The acoustic properties of the stimulus were checked with a sound analyzer before proceeding with the experiment. To confirm that the stimulus met the desired specifications, the synthetic sound was acoustically analyzed in Praat . The final recordings were analyzed using a signal processing tool developed with MATLAB 6.0 for Windows and the peaks were named in the same way as those of cABRs .
2.1
Patients
Twenty subjects institutionally diagnosed as dyslexics were studied. All subjects were Greek speakers with Greek as their first language without a history of brain damage, language problems, psychiatric symptoms, or visual problems. Eight were males and 12 were females. They were between 18 and 23 years of age, with a mean age of 20 years. The control group consisted of 20 age- (19–22 years, mean 20.5), sex-, education-, hearing sensitivity- and IQ-matched healthy young adults. The IQ was estimated in our Institutions, from four index scores (verbal comprehension, perceptual reasoning, working memory, and processing speed) from the Wechsler Adult Intelligence Scale (WAIS IV) . Subjects with an IQ below 85 were excluded.
All subjects underwent a thorough audiological examination. Audiometric tests included speech reception threshold testing, pure-tone audiometry, speech discrimination testing, and tympanometry.
2.2
Broad band ‘fast’ and complex ‘long’ stimuli
In all subjects, the ABRs were recorded with the person in the supine position in a quiet room. All tests were performed in all subjects using auditory stimuli of broad spectrum (click) with an intensity of 80 dB nHL, repetition rate 11.29/s and frequency filters 100–3000 Hz. The parameters evaluated were the latencies for the waves I to V, and the interpeak latencies I–III, III–V, and I–V .
Since broad spectrum stimuli evoke brainstem potentials by the synchronous firing of more individual neural units, these potentials are very clear . However a fast-onset stimulus contains transient acoustic energy only at various frequencies, with frequency specificity in dyslexia. In order to eliminate such transients and obtain a good stimulus specificity, the cABRs of the subjects were also tested, using ‘complex’ acoustic stimuli with sufficiently long onset time and wider frequency band .
2.3
ABR acquisition
A PC-based stimulus delivery system controlled the time of delivery, stimulus sequence, and stimulus intensity, and triggered the PC-based evoked potential averaging system. All electrode impedances were < 5 kΩ. A MATLAB for Windows-based software tool identified and marked the onset, offset, and peak latencies and a trained observer resolved discrepancies on the PC-output of the system signal. Reproducibility of the results was always achieved. The cABRs were collected according to the method described before, in response to a click (0.1 ms) and randomly presented alternating polarities of ‘ba’. Alternating polarities were added together to isolate the neural response from that of the cochlear microphonic. cABRs were differentially recorded from Cz-to-ipsilateral earlobe, with the forehead as ground. The sampling rate was 20,000 Hz. Responses were bandpass filtered on-line from 100 to 2000 Hz. Sweeps with noise levels that exceeded ± 30 μV were rejected from the average. Three repetitions of 1000 sweeps each were collected in response to the click as well as for each polarity of ‘ba’. The click stimuli were presented at 80 dB SPL with an interstimulus interval (ISI) of 32 ms; the recording window was 20 ms, including a 10 ms prestimulus period .
The ‘ba’ stimuli were presented at 80 dB SPL with an ISI of 51 ms; the recording window was 60 ms, including a 10 ms prestimulus period . The latencies of the click-evoked waves I–V and the negative peaks following the stimulus ‘ba’, marked A and C, were compared with normative values. Negative peak A is the onset of the response to the stimulus and peak C is the initial part of the frequency following the onset of the response ( Fig. 1 ). We used the latencies of waves A and C as well as the interpeak latencies A–C, and compared them between normal and dyslexic subjects. On this basis, a multiparametric analysis (filtering, peak identification, and latency calculation) through a PC-based signal processing tool implemented with a MATLAB tool of the recordings of the examined normal and dyslexic cases was performed. The possible significant differences between the two methods in both groups were assessed using Student’s t-test for unequal samples .
