2003 IFA Congress: Montreal, Canada

Cerebral Lateraization of Speech Processing in Adult and Child Stutterers: Near Infrared Spectroscopy and MEG Study

Koichi Mori1, Yutaka Sato1, Emi Ozawa1 and Satoshi Imaizump2
1National Rehabilitation Center for Persons with Disabilities, 4-] Namiki, Tokorozawa, Saitama, 359-8555, Japan
2Hiroshima Prefectural College of Health Sciences,I -I Gakuen, M ihara, Hiroshima 723-0053, Japan


Cerebral lateralization of speech processing in stutterers were assessed with noninvasive brain imaging techniques, magnetoencephalography and multichannel near infrared spectroscopy (NIRS), with which neuromagnetic and hemodynamic responses, respectively, were recorded to analysis-synthesized prosodic and phonemic minimal contrast word trains. Adult stutterers did not show normal leftward dominance for the phonemic contrast with either method. Children underwent only NIRS sessions, with results similar to those of adults, which indicates that the cerebral dominance in processing heard speech is in disarray even in school-age stutterers. The NIRS method may be useful in screening young stutterers and in elucidating neural correlates of stuttering.

  1. Introduction
Although the cause(s) of stuttering is not known, failed establishment of cerebral dominance for speech processing has been demonstrated in adult stutterers with various brain mapping techniques including positron emission tomography (PET) (Braun et al., 1997; Fox et al., 1996; Wu et al., 1995) and magnetoencephalography (MEG) (Salmelin et al., 1998). Braun et al. (1997) reported that rCBF patterns in stuttering subjects differed markedly during the formulation and expression of language, failing to demonstrate the left hemispheric lateralization that is typically observed in normal controls. In another PET study (Wu et al., 1995), stutterers showed significant decreases in regional glucose metabolism in Broca's area, Wernicke's area and frontal pole in a reading aloud (stuttering) condition compared to a non stuttering condition. Fox et al. (1996) characterized developmental stutterering not only with extensive hyperactivity of cerebral and cerebellar motor system, but also with right dominance in the cerebral motor system. They deduced that the surprising levels of deactivation in the temporal lobe correspond to the stutterers’ lack of normal monitoring of their speech (Ingham, 2001). Salmelin and colleagues indicated that the functional organization of the stutterers’ auditory cortex was fundamentally different from that of normal subjects. In fluent speakers, the left auditory cortex is more sensitive to the side of stimulation (right versus left ear), whereas the right auditory cortex is more sensitive in stutterers. They concluded that the interhemispheric balance is more unstable in stutterers and this might lead to disturbances in the control of speech (Salmelin et al., 1998).

While these MEG and PET studies -generally suggest that stuttering and nonstuttering speakers differ in the lateralization of various cortical processes, the crucial question for elucidating the possible causal relationship between stuttering and brain lateralization is how the development of speech dominance in children is affected by stuttering, or vice versa. Because conventional functional brain mapping techniques are not well-suited for children, due to safety concerns and/or requirement for rigorous restraint, few studies if any have been conducted in stuttering children. The aim of this study is to investigate the functional laterality in adult and child stutterers during auditory language processing. The present study employed multichannel near infrared spectroscopy (NIRS), which is optical, non-invasive monitoring system of cerebral hemodynamics that can be easily used with children and infants for assessing cerebral lateralization for speech. In order to validate the NIRS method, MEG was also used for adult stutters with the same set of stimuli.

  1. Methods
2.1 MEG experiment

Subjects and procedures: Three adult stutterers (3 male, age range 22-32 years) and ten fluent speakers (Imaizumi et al., 1998) participated in the MEG experiment. Written informed consent was obtained from each subject in accordance with the Declaration of Human Rights, Helsinki, 1975. All subjects were right-handed, native speakers of Japanese. With a 122-channel whole head SQUID magnetometer (Neuromag Ltd., Finland), the elicited magnetic fields (MFs) were measured.

Stimuli: Three different forms of the Japanese verb /iku/ (meaning "to go") were produced with a synthesis by analysis system (ASL, Kay Elementrics Corp., USA) based on a speech signal recorded by a male adult (Imaizumi et al, 1998). By changing the vocal pitch contour and the formant frequencies, (A) past declarative /itta/ ("went"), (B) interrogative /itta?/ ("went?"), and (C) imperative /ittte/ ("Go away"), were synthesized. Only the final syllable was changed among the three words. Particular combinations of two of the three words formed phoneme (/itta/ and /itte/) or pitch (/itta/ and /itta?/) contrasts. An oddball-like paradigm was adopted. One of the three words was presented as the high frequency standard stimuli and the others as deviant rare stimuli at the frequency of one per 6 standard presentations. The inter-stimulus interval was randomly varied from 0.9 to 1.0 s. The stimuli were presented 800 times per each session in a randomized order, through a pair of tube earphones (EAR-TONE 3A) at the comfortable sound level.

Five sessions were performed: three ignore and two attention conditions. The three ignore sessions were carried out with the three different standards /itta/, /itta?/ and /itte/, under the instruction to ignore the speech stimuli. Under the attention conditions, /itta/ was presented as the standard and the subjects were instructed to count the number of times either /itta?/ or /itte/ was heard. The order of the sessions was counterbalanced across subjects.

Data analysis: The difference MFS or mismatch magnetic fields (MMFS), corresponding to the mismatch negativity (MMN) in evoked potentials, were calculated by subtracting the MFS that were elicited when a word was the standard from those when the same word was a deviant (Imaizumi et al., 1998). Application of a single-current dipole model to the MMFS estimated the location, latency, and moment (Q) of their equivalent current dipoles (ECDs). The model employed was considered to adequately represent the measured MFs if the goodness-of-t (g) between the data and the model was > 0.7. The ECD latency was determined as the time corresponding to the peak value of Q between 150 and 200 ms after the onset of the final syllables.

The effects of the group (stutterers, fluent speakers), the contrast condition (prosody, phoneme, both) and side (left, right) on the ECD moments were analyzed by a three-way analysis of variance (ANOVA).

2.2 NIRS experiment

Subjects and procedures: Ten stuttering adults (10 male, age range 18-44 years) and five school-age stuttering children (3 male, 2 female, age range 6-12 years) who stutter participated in the NIRS experiment. All subjects were right-handed as assessed by Edinburgh Handedness Inventory (Oldfield, 1971) and native speakers of Japanese. Subjects were recruited in Hospital of National Rehabilitation Center for Persons with Disabilities (NRCD), Japan, and a self-help group for stuttering. Written informed consent was obtained before the experiment. The research was approved by the ethical committee of NRCD.


Figure l. NIRS measurement system [Left panel] Optical probes were placed on the scalp with thermoplastic shells. All probes were connected to the data acquisition system with flexible optical fibers. [Right panel] Near infrared light arrives at the brain tissue through the skin and skull, and returns to the detection probe by scattering.

Recordings of the changes in hemoglobin (Hb) concentrations in the bilateral temporal areas were made with a 24-channel NIRS system (ETG-l0O, Hitachi Medical Co., Japan), which uses continuous near infrared lasers at two wavelengths modulated at different frequencies and detected with lock-in amplifiers (Watanabe et al., 1996). The recording channels resided in the optical path in the brain between the nearest pairs of incident and detection probes which were separated by 3 cm on the scalp surface. Five incident and four detection probes arranged in a 3 x 3 square lattice were placed on each lateral side of the head, which made the total number of recording channels 12 on either side (Fig. 1). After the recording, the tip positions of optical probes were measured with a 3D digitizer (Polhemus, Vermont, USA). The coordinates of the centers of the nearest incident and detection probe pairs were calculated, which served as the lateral pointers to the actual centers of respective recording volumes in the brain. The centers of recording sites were identified by superimposing the above coordinates onto Tl-weighted parasagittal MR brain images for each adult subject. The channels nearest to the lateral end of the border between the transverse temporal gyrus (TTG) and the planum temporale (PT) in a parasagittal projection were presumed to be in the auditory area (Furuya & Mori, 2003; Minagawa-Kawai et al., 2002). This procedure selected the recording channels whose centers were within the 1.5 cm radius of the TTG-PT border, and thus should contain the signals in the auditory cortex due to the spread of the laser in the brain tissue (Yamashita et al., 1996). Since it was difficult to acquire MR brain images of some child subjects without anesthesia, the positions of optical probes were recorded with either a 3D digitizer or a digital camera for identification of approximate recording areas.

Stimuli:Two sessions were performed using the same word stimuli as for MEG experiments in a block design paradigm. In the phoneme contrast session, the baseline block contained only /itta/ which was repeated approximately every second, whereas the test block consisted of /itta/ and /itte/ presented in a pseudo-random order with the equal probabilities at the same rate as in the baseline block. Both blocks lasted for 20 s respectively, and presented alternately at least five times. The pitch contrast session was the same as the phoneme contrast session except for the presentation of the /itta/ and /itta?/ combination in the test block. The order of two sessions was counterbalanced among subjects. Stimuli were presented at a comfortable level (60-70 dB SPL) via insert earphones (EAR-TONE 3A) to adults and a loudspeaker (il5, TANNOY) to children.

Data analysis: The Hb contrast data were sampled at 10 Hz and smoothed with a 5 s moving average. The concentration of total Hb during the test block was averaged after excluding the blocks with artifacts in each session. The maximal total Hb change was calculated against the 10 s pre-test baseline period in each channel. In order to assess cerebral lateralization, laterality index, Ll=(L-R)/ (L+R), was calculated from the peaks of the maximal total Hb responses in the left (L) and the right (R) auditory areas. Ll could range from -1 to l, with a positive value indicating left dominance. Lls were compared between the two conditions (Mann-Whitney U-test).

Within subject analysis was also performed. Without averaging over repeated blocks, the left and right peaks of Hb concentration changes were obtained for individual test blocks, for which Lls were calculated and pooled for comparison between the two contrast conditions within each subject (Mann-Whitney U-test).

  1. Results


The average peak moment (Q) of the ECDs of MMFS of the stutterers was significantly smaller than that of the nonstutterers (p < 0.01, ANOVA) in the ignore conditions. The main effects for other factors and interactions were not significant in the ignore conditions.

Figure 2 shows the Q values of the ECDs under the attention conditions. Only the group effect was significant (stutterers < fluent speakers; p < 0.01, ANOVA). No significant differences in ECD locations and latencies were found between groups or conditions.


Figure 2. Dipole moments for phoneme and pitch contrasts under the attention conditions. Filled bars: left, white bars: right dipole moments in the auditory area. Error bars: one standard error.


Figure 3 shows NIRS responses in an adult stutterer. The peak concentration change in total Hb was larger in the right auditory channel than in the left in response to the presentation of the phoneme contrast. The left-right preponderance in the response to the pitch contrast was opposite to that to the phoneme contrast. This left-right pattern of activation is reversed from the normal control (Furuya & Mori, 2003; Figure. 4 left).

Because the stutterers showed much wider variation in the relationship between the LIs to pitch and phoneme contrasts than the control, neither children or adult stutterers showed any significant difference in L1 between the two contrast conditions as a group.(Figure 4).


Figure 3. Hb responses evoked by phoneme and pitch contrasts. [Upper Panel] The probe locations (black circles) and the center of the presumed measurement channels (white circles) are superimposed onto the left and right parasagittal MR images. Anterior is positive in the scales above the images, with 0 mm at the pre-auricular points. The lateral posterior borders of the Heschl gyrus are labeled "A". The channels with themaximal responses on respective sides are shown with gray circles. [Lower Panel] The averaged Hb concentration changes in response to the two contrasts (Phoneme: black arrows, upper graph; Pitch: light arrows, lower graph) recorded at the same left and right channels shown in the upper panel.


Figure 4. Laterality indices for the phoneme and pitch contrasts. [Left Panel] The Lls for the phoneme (white boxes) and pitch (gray boxes ) contrasts of adult nonstutters ( F uruya & Mori, 2003) and stutterers. [Right Panel] School-age stutterers as a group did not show a significant difference between the two conditions. The nonstutter control group consisted of 3 -5 years old children (Sato et al., 2003). Boxes: the quartiles. Bars in the boxes: the medians. Hinges: the ranges. *: a significant difference of LIs between the phoneme and pitch contrast (p < 0.01). N.S.  not significant (p > 0.05).

Within-subject analysis revealed that none of the stutterers showed a significant difference between pitch and phoneme Lls, except two adults (the results of one of whom are presented in Fig.2) and one child showing rightward LI for the phoneme contrast. This is markedly different from the right-handed control where 85% of adults showed significantly more leftward Lls for phoneme than for pitch, with the remaining 15% showing no significant difference between the two contrast conditions (Furuya & Mon, 2003) (Figure 5).


Figure 5. Within-sabject analysis for Us between phoneme and pitch contrasts

Cerebral dominance as percentage of subjects is plotted for each group with the number of subjects in each bar. Oblique hatched bars: a significantly more positive L1 in the phoneme than in the pitch contrast session, indicating the left side dominance in the phoneme processing. Grey bars: no significant difference between the contrast condition. White bars: a significant negative shift of L1 in the phoneme contrast condition relative to the pitch.

  1. Discussion
We have demonstrated the differences between developmental stutterers and nonstutterers in cerebral lateralization in the auditory processing of prosodic and phonemic contrasts by using NIRS and MEG. Abnormal functional lateralization in speech processing has been already demonstrated with MEG (Salmelin et al., 1998), with which our MEG results are in line. Using the same stimuli as MEG, we investigated adult and school-age stutterers with NIRS, with the adult subjects showing similar results to those with MEG, but with a hint of a higher sensitivity of NIRS than MEG. Both adult and school-age children who stuttered showed abnormality in the lateralization of phoneme contrast processing, which is consistent to and extends the previous finding of different functional interhemispheric asymmetry of auditory language processing between adult stutterers and nonstutterers (Fox et al., 1996; Salmelin et al., 1998; Curry & Gregory, 1969). Because all the stutterers in our study demonstrated abnormality in functional lateralization of speech processing, the diagnostic sensitivity for stuttering of our method is 100%, while its specificity is 85% according to the results from the adult control (Fig. 3). Our results in children indicate that the abnormality in the auditory functional lateralization starts at least as early as the elementary school age.

The etiology of developmental stuttering is still unknown, although various abnormalities have been postulated as its causes; the laryngeal control (Conture, 1986), the motor systems controlling the speech organs (Fox et al., 1996), the Broca’s area and speech planning (Wu et al., 1995), neural processing sequence among the motor and the Broca’s areas (Salmelin et al., 2000). Although stuttering refers to a speech motor dysfunction, previous studies of functional lateralization in stutterers using various methods (Curry and Gregory, 1969; Hall and J erger, 1978; Fox et al., 1996; Salmelin et al., 1998; this study) have found less-than-norrnal left-side dominance for linguistic processing and suggest the significant involvement of the auditory system in the disorder. The fact that fluency can be induced with delayed auditory feedback (DAF, Lee, 1950) or white noise that masks self-monitoring of own speech (Cherry & Sayers, 1956) also suggests the crucial role of the auditory system in stuttering.

The abnormal functional lateralization of language processing that has been so far demonstrated in the adult auditory brain, however, might be a result rather than a cause of stuttering, due to the possibility of compensatory plasticity during development with long-standing stuttering. The present results suggests either (1) that the abnormality is causal to stuttering, or (2) that stuttering persisting for only several years is enough to reset or even reverse the functional lateralization of the auditory area that should have been already established at the age of one year (Sato et al., 2003).

This long-standing issue can only be resolved by investigating younger stutterers than those studied here. Towards this end, we have chosen a novel functional brain mapping technique using multichannel NIRS, and a task that does not require active participation on the subjects. With this paradigm, the development of functional lateralization of speech processing in the auditory area can be monitored in infants (Sato et al. 2003). Behavioral tests requiring intensive attention, like dichotic listening tests, may not be possible with young children at the age of the highest risk of suffering from stuttering. Tasks requiring listening only Without reading or writing allow the study of illiterate and preliterate children (Ahmad et al., 2003). Even speech tasks, as employed in previous adult studies (Fox et al., 1996; Salmelin et al., 1998), may not be reliably performed by younger stutterers.

Conventional neuroimaging techniques either do not have enough resolution (eg., evoked potentials), are not safe enough (PET and single photon emittion computerized tomography, SPECT), or are too restrictive (PET, SPECT, functional MRI) to be used for young stutterers. N IRS allows noninvasive measurement of human brain functions under a variety of conditions with little restraint of the subject (Kennan et al., 2002; Zaramella et al., 2001), and has a reasonable resolution due to the limited spread of near infrared light in the tissue (Yamashita et al., 1996), unlike evoked potentials.

As the NIRS method is best suited among the available brain mapping techniques for studying the lateralization of cortical auditory functions in children and infants, it may be useful for elucidating neural correlates of stuttering, and even evaluating and diagnosing stuttering in infants in the future.

  1. Conclusion
NIRS and MEG were used to investigate the cerebral dominance of the auditory language processing in adult and child stutterers. Analysis-synthesized Japanese word stimuli with phoneme (/itta/ and /itte/) and pitch (/itta/ and /itta?/) contrasts were used. With NIRS measurement, we calculated the laterality index (LI) from the peaks of the left and right total Hb responses to each contrast in the auditory area. The results showed that there were no significant differences in L1 between phoneme and pitch contrast responses in stutterers. Within-subject analysis showed no subject with a significant leftward shift of the L1 in phoneme condition relative to pitch condition, although there were a few with a reversed laterality in comparison to the normal control. Similar trends were observed in the MEG study. These results confirm that stuttering is closely correlated to the abnormality of the cerebral dominance for processing heard speech, even in school-age children. The NIRS method used in this study has a potential clinical application as an objective test of stuttering because it enables evaluation of individual cerebral laterality. It also may help elucidating how the cerebral lateralization is related to stuttering.

Ahmad, Z., Balsamo, L. M., Sachs, B. C., et al. (2003). Auditory comprehension of language in young children: Neural networks identified with flV1RI. Neurology, 27; 60, 1598-1605.

Braun, A. R., Varga, M., Stager, S., et al. (1997). Altered patterns of cerebral activity during speech and language production in developmental stuttering. An H2150 positron emission tomography study. Brain, 120 ( Pt 5), 761-784.

Cherry, C., & Sayers, B. (1956). Experiments upon the total inhibition of stammeing by external control, and some clinical results. J. Psychosom. Res., 1, 233-246.

Conture, E. G., Rothenberg, M., Molitor R. D. (1986) Electroglottographic observations of young stutterer’ ï¬ uency. Journal of Speech and Hearing Research, 29, 384-393.

Curry, F. W., & Gregory, H. H. (1969). The performance of stutterers on dichotic listening tasks thought to reflect cerebral dorninance.Journal. Speech and Hearing Research, 12, 73-82.

Fox, P. T., Ingham, R. J ., Ingham, J . C., et al. (1996). A PET study of the neural systems of stuttering. Nature, 382(6587), 158-161.

Furuya, 1., & Mori, K. (2003). [Cerebral lateralization in spoken language processing measured by multi-channel near-infrared spectroscopy (NIRS)]. No To Shinkei, 55(3), 226-231. in Japanese.

Hall, J. W., & Jerger, J . (1978). Central auditory function in stutterers. Journal of Speech and Hearing Research., 21, 324-337. â_˜

lmaizumi, S., Mori, K., Kiritani, S., et al. (1998). Task-dependent laterality for cue decoding during spoken language processing. Neuroreport, 9, 899-903.

Ingham, R. J. (2001). Brain imaging studies of developmental stuttering. Journal of Commuication. Disorders, 34, 493-516.

Kennan, R. P., Kim, D., Maki, A., et al. (2002). Non-invasive assessment of language lateralization by transcranial near infrared optical topography and functional MRI. Hum. Brain Mapp., 16(3), 183-189.

Lee, B. S. (1950). Effects of Delayed Speech Feedback. J. Acoust. Soc. Am, 22, 824-826.

Minagawa-Kawai, Y., Mori, K., Furuya, 1., et al. (2002). Assessing cerebral representations of short and long vowel categories by NIRS. Neuroreport, 13(5), 16581-16584.

Oldfield, R. C. (1971) The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia, 9, 97-113. ,

Salmelin, R., Schnitzler, A., Schrnitz, F., et al. (1998). Functional organization of the auditory cortex is different in stutterers and fluent speakers. Neuroreport, 9, 2225-2229.

Salmelin, R., Schnitzler, A., Schmitz, F., et al. (2000). Single word reading in developmental stutterers and fluent speakers. Brain, 123 ( Pt 6), 1184-1202.

Sato, Y., Mori, K., Furuya, 1., et a1. (2003). [Developmental changes in cerebral lateralization during speech processing measured by near infrared spectroscopy.] Onseigengoigaku, 44, 165-171. in Japanese.

Watanabe, E., Yamashita, Y., Maki, A., et al. (1996). Non-invasive functional mapping with multi- channel near infra-red spectroscopic topography in humans. Neuroscience Letters, 16, 41-44.

Wu, J . C., Maguire, G., Riley, G., et al. (1995). A positron emission tomography [18F]deoXyg1ucose study of developmental stuttering. Neuroreport, 6, 501-505.

Yamashita, Y., Maki, A., and Koizumi, H. (1996). Near-infrared topographic measurement system: Imaging of absorbers localized in a scattering medium. Rev. Sci. Instrum., 67(3), 730-732.

Zaramella, P., Freato, F., Amigoni, A., et a1. (2001) Brain auditory activation measured by near-

infrared spectroscopy (NIRS) in neonates. Pediatric. Res., 49, 213-219.


In preparation for the 2018 World Congress the IFA is implementing Japanese translations of some pages on the site. Choosing Japanese below to see these translations.

Not all pages are translated, but you can use Google translate to see a machine translation using the switch below

Google Translate

Follow the Joint World Congress