Identification of binaural integration deficits in children with the Competing Words Subtest: Standard score versus interaural asymmetry

Abstract

The Competing Words Subtest is a commonly used dichotic listening test for assessing binaural integration in children suspected of having an auditory processing disorder. In 124 children, standard scores from the subtest suggested a binaural integration deficit in 23% of the children tested. Because standard scores are derived from the combined scores of both ears during the test, children with normal performance in one ear and weak performance in the other ear may be overlooked. For these children, a measure of interaural asymmetry may be a more sensitive indicator of a binaural integration deficit. When an age-appropriate criterion for interaural asymmetry from the Competing Words Subtest was used, the incidence of a binaural integration deficit increased to 51% of the children tested. Four typical patterns of dichotic listening performance were identified based on results from the two scoring techniques.

Sumario

La prueba competitiva con palabras es una prueba dicótica comúnmente utilizada para evaluar integración binaural en niños en los que se sospechan alteraciones de procesamiento central. En una muestra de 124 niños las calificaciones estándar sugerían un déficit en la integración bianural del 23%. Debido a que las calificaciones estándar son derivadas de puntuaciones combinadas de ambos oídos, los niños con un desempeño normal en un oído y pobre en el otro, pueden pasar desapercibidos. La medición de la asimetría interaural para estos niños puede ser un indicador más sensible de un déficit de integración binaural. Cuando se utilizó un criterio apropiado, por edades para evaluar asimetría interaural de la prueba competitiva con palabras, la incidencia de asimetría en la integración binaural se elevó a un 51% en los niños evaluados. Con base en los resultados al utilizar las dos técnicas de calificación, se identificaron cuatro patrones de desempeño en las pruebas dicóticas.

• Auditory processing disorder
• Dichotic listening
• Interaural asymmetry
• Children

Acronyms
SCAN	=	Screening Test for Auditory Processing Disorders
SCAN-C	=	Test for Auditory Processing Disorders in Children – Revised
ASHA	=	American Speech Language and Hearing Association
APD	=	auditory processing disorder
PTA	=	pure-tone average
SS	=	standard score
IA	=	interaural asymmetry

Moncrieff, D. W. 2006. Identification of binaural integration deficits in children with the Competing Words Subtest: Standard score versus interaural asymmetry. Int J Audiol, 45, 200–210.

Due to a production error, the above discussion paper was published without its associated figures and legends. The corrected paper is reproduced on the following pages.

An ASHA Task Force (1996), recommended that a comprehensive assessment for auditory processing disorder (APD) must include behavioral tests for several auditory processes, including performance decrements with competing acoustic signals. For many years, tests of binaural integration have been used to assess a listener's ability to process competing auditory information presented simultaneously to the two ears. The standard test of binaural integration is the dichotic listening test administered so that the listener is asked to repeat all information following presentation to both the right and left ears (Musiek & Pinheiro, 1985; Bellis, 2003). Because more elements are presented than can be consciously analyzed with ease, the listener must select or attend to some stimuli at the expense of others, usually leading to better performance in one ear than in the other. According to one theory, the direction and degree of interaural asymmetry is thought to reflect laterality differences between the two hemispheres of the brain as activated by each ear. This structural model of dichotic listening states that cortical processing favors input of verbal material from the ear that is contralateral to the language-dominant hemisphere, resulting in better performance in the right ear for most individuals (Kimura, 1967). An alternative attentional model suggests that when listeners direct attention to one side or the other, identification of verbal material presented to that side will be improved and results may shift laterality in that direction (Kinsbourne, 1970; Hiscock & Beckie, 1993). Forced attention during dichotic listening has been shown to enhance each ear's individual performance, but directing attention first to the right and then to the left did not result in an overall change in children's ear advantages for digits and words (Obrzut et al, 1986). Another factor that has been shown to affect performance directly and interact with attention is handedness (Mondor, 1994). A right-handed listener may produce the typical right-ear advantage (REA) during dichotic listening, but it would also be normal for some left-handed listeners to produce a left-ear advantage (LEA) (Hugdahl et al, 1997). Because handedness may be related to structural differences in the brains of individuals being tested, the direction of laterality may vary and both directions may be normal. Whether the dichotic listening test is performed in the free recall mode with no specific direction of attention or in a forced or directed attention mode of presentation, the degree of interaural asymmetry reflected by the overall performance difference between the two ears may reflect binaural integration performance better than the direction of laterality.

There are four patterns of responses that typically occur during the dichotic listening task. Normal listeners respond with strong performance from both ears with one ear performing slightly more poorly than the other ear, resulting in a small interaural asymmetry (Hugdahl, 1995). Other listeners may respond with a small interaural asymmetry also, but demonstrate poor performance in both ears. This second pattern has been documented in patients with brainstem and cortical lesions (for review, see Musiek & Pinheiro, 1985). In children with no known lesions, similarly poor performance in both ears could be due to problems with binaural integration as well as to other factors, including language disorder, attention deficit disorder, short-term memory problems, poor motivation, fatigue, or limited cognitive abilities. In a third pattern, the listener's ear that is contralateral to the language-dominant hemisphere performs normally and the other ear performs poorly, resulting in a large interaural asymmetry. This pattern has suggested limited myelination of the corpus callosum leading to poor interhemispheric transfer of linguistic information in both children and adults (Jerger, Johnson & Loiselle, 1988; Jerger et al, 2000). Poor performance in one ear is sometimes referred to as a left-ear deficit and is also linked to an auditory processing disorder known as an integration deficit (Keith & Engineer, 1991; Bellis & Ferre, 1999; Jerger et al, 1999; Musiek, 1999; Moncrieff & Musiek, 2000). Reduced performance in only one ear during a single dichotic listening test could be due to a failure of the listener to direct attention toward that ear, but if the pattern occurs similarly across multiple dichotic listening tests, it is less likely due to attention or other factors. Some children demonstrate both the second and third patterns of difficulty with poor performance in both ears together with a large interaural asymmetry. This combined difficulty represents the fourth pattern which suggests that there may be multiple factors involved in the results from dichotic listening tests.

A commercially available dichotic listening test for assessing binaural integration in children that has been in widespread use since its original development as the SCAN in 1986 is the Competing Words Subtest of the SCAN-C (Keith, 2000). It is the only single word dichotic listening test that has been standardized for children. Instructions with the revised test direct clinicians to combine separate ear scores into an overall raw score and to then convert the raw score into a standard score for interpretation of the results. Standard scores that fall below 7 are considered borderline if they exceed 4 and disordered if they are at 4 or below. One potential problem with this standard scoring method is that children whose performance is poor in both ears are more likely to be identified as having a deficit than are children with poor performance in only one ear. When the raw scores from the two ears are combined for children with reduced performance in only one ear, the score of the stronger ear may be sufficient to pull the overall score into a normal range of performance. In this way, children who have problems with the interhemispheric transfer of linguistic information through the corpus callosum may be overlooked.

During standardization of the SCAN and SCAN-C, normal values for interaural asymmetry across the Competing Words Subtest were not reported. Instead, prevalence of interaural asymmetry during each of the two listening conditions was established and persons administering the test were directed to compare a child's performance with those values during each listening condition. If a child's performance occurred in only 2% of the children evaluated during standardization of the test, the performance was to be considered abnormal. Since nearly 10% of the standardization sample was known to have a variety of difficulties that could impair auditory processing performance, this criterion could potentially identify only children with severe binaural integration deficits. Furthermore, this measurement fails to account for the overall performance difference between the two ears and may be subject to factors related to attention or direction of hemispheric lateralization. For example, a 10- to 11-year-old child's results would be considered typical with a maximum overall laterality of 20%, having demonstrated a right-ear advantage of 4 in the right ear first condition, and a right-ear advantage of 2 in the left ear first condition, thereby producing an overall ear advantage of 6 out of 30 possible correctly identified words. To be considered abnormal for possible APD diagnosis on the basis of an ear advantage that occurred in only 2% of the standardization sample, the same age child could demonstrate an overall laterality of as little as 17%, with a right-ear advantage of 8 in the right ear first condition and a left-ear advantage of 3 in the second condition. It could be argued that in the second case, the child demonstrated improvement in performance in the nondominant ear during the second half of the test when directed to respond from that ear first as predicted by Kinsbourne (1970). In this case, an overall measure of interaural asymmetry would be a more appropriate measure of performance than the measure taken during either of the individual directed conditions, as suggested by Obrzut et al (1986).

Another scenario that makes this scoring method problematic is that some left-handed children may demonstrate a normal degree of interaural asymmetry, but it would be reflected as a left-ear advantage if they lateralize language in the right hemisphere of the brain. For a left-handed child, a left-ear advantage of 4 in the left ear first condition would result in a referral because that pattern occurred in less than 2% of the standardization sample. In this case, the measure of overall interaural asymmetry could be as little as 13%. Because this scoring method is complicated and seems vulnerable to misinterpretation, it would simplify the process if an overall measure of interaural asymmetry could be used to reflect the difference between a child's stronger and weaker ears, regardless of which ear is the dominant one. If directing the response first to one side and then to the other should not alter the overall measure, then the final difference score between the two ears will provide an accurate measure of both interaural asymmetry and laterality during a dichotic words test.

Interaural asymmetry is usually higher with words than with digits due to the higher verbal workload involved in identifying targets from an open set of single syllable words as contrasted to the identification of a closed set of single syllable digits from 1 to 10 (Porter & Berlin, 1975). Early in development, reduced performance in both ears together with a larger interaural asymmetry is normal for children. As children mature, performance in each ear should improve with greater increases in performance in the weaker ear relative to the stronger ear. Therefore, interaural asymmetry reduces as performance in the weaker ear improves to levels closer to the performance in the stronger ear. As a result, it is essential that normal values for individual ears and for interaural asymmetry be established for children's performance with dichotic words so that results can be clinically interpreted. With dichotic digits, interaural asymmetry rarely exceeds 10% in very young children (Harper & Kraft, 1994) and typically shrinks to approximately 2% by age 11 (Bellis, 2003). With words, average measures of interaural asymmetry in two early studies with children were never greater than 19% (Thomson, 1976; Obruzut et al, 1986). In a more recent study comparing 11-year-old right-handed dyslexic children to normal controls, a Competing Words Subtest interaural asymmetry greater than 10% was related to poorer reading decoding skills and differentiated the two groups of children (Moncrieff & Musiek, 2000). Results from the standardization of the SCAN-C (Keith, 2000) failed to report values for interaural asymmetry, but average performance for both ears improved by nearly 5% for each increase in age of one year. A conservative approach would be to suggest that the youngest children at ages 5 to 6 years should have demonstrated no greater than a 40% difference between ears. With maturity, that value should have improved approximately 10% for every two years of development. Therefore, interaural asymmetry values should have been no greater than 30% for children ages 7 to 8, 20% for children ages 9 to 10, and 10% for children ages 11 to 12.

The primary goal of the study was to first measure the incidence of a binaural integration deficit in children as determined by standard scores and to then compare it to the incidence of a deficit based on measures of interaural asymmetry. A binaural integration deficit was initially determined by either a borderline or disordered standard score on the Competing Words Subtest and was then determined by the presence of a larger-than-normal measure of interaural asymmetry using the conservative values suggested by results from the standardization of the SCAN-C. It was predicted that a combination of both interaural asymmetry and standard score would identify more children with binaural integration deficits than standard score alone. Because any listener may produce a large interaural asymmetry on a single test, interaural asymmetry measures were validated across a number of other dichotic listening measures for a subgroup of children involved in this study. In addition to measuring the incidence of a binaural integration deficit on the basis of these different scoring criteria, another goal of the study was to separate children into subgroups, using standard scores and interaural asymmetry to create four different patterns of performance similar to the patterns described from research on patients with known lesions of the central auditory nervous system.

Method

Subjects

A total of 124 children were included in the study. Forty of the children were recruited for participation in studies focused on auditory processing by distribution of a flier throughout locations in Gainesville, Florida and through an insert mailed with the regional utility bill. Parents contacted the university and were informed that the children would be participating in a variety of studies of auditory processing skills. These children were evaluated at the auditory processing laboratory at the University of Florida. Information for the remaining 84 children came from several sources. Information for 72 of the children was obtained by a retrospective review of clinical files for assessments made during the previous three years at the University of Florida Speech and Hearing Clinic and at the Department of Communicative Disorders at the University of Florida Shands Hospital. Another twelve children were referred for clinical assessment to the University of Florida Speech and Hearing Clinic and were evaluated in the auditory processing laboratory. The distribution of gender and age across the children tested is shown in Figure 1.

Figure 1.  Number of females and males at each age involved in the study.

Procedure

Children were seated in a sound-attenuated room. Test material was delivered from a compact disc through a clinical audiometer driving either supra-aural or insert earphones. The pure-tone averages (PTAs) of hearing thresholds at 500, 1000, and 2000 Hz were measured and recorded according to standard clinical procedures. Dichotic listening tests were administered according to standard procedures included in the instructions for each of the tests. Within the 124 children, 26 of the children referred for clinical evaluation were tested in 1999 and 2000 with the Competing Words Subtest of the SCAN and the remaining children were all tested with the SCAN-C. Other tests used to further evaluate some of the children included the Staggered Spondaic Words Test (SSW) (Katz, 1968) the Dichotic Double Digits (DDT) (Musiek, 1983), the Randomized Dichotic Digits Test (RDDT) (Strouse & Wilson, 1999) and the Dichotic CVs Test (DCV) (Hugdahl et al, 1990). Scores were recorded for number and percent correct in the right ear and the left ear across all dichotic listening tests. The standard score was determined from the combined ear scores measured during the Competing Words Subtest. Thirty-nine of the children were also assessed for handedness (Annett, 1970) by being asked to demonstrate their use of several everyday objects.

Results

Binaural integration deficits based on standard scores

All children produced pure-tone averages in both ears that were within normal limits (≤25 dB HL). Standard score results ranged from 1 to 18. A Kolmogorov-Smirnov test with Lilliefors significance correction showed that standard scores were not normally distributed in this group of children. It was not unexpected that these results were not normally distributed since 68% of the children had been referred for a clinical evaluation for APD. Non-normal distributions often occur when there are outlier or extreme scores. A box plot of the standard scores produced by children in this study is displayed in Figure 2. Median performance is indicated by the black bar inside the box and the box itself represents the interquartile range of results. Outliers are denoted by circles and represent values that are between 1.5 and 3 box lengths from the edge of the box. One child's standard score of 1 represented an outlier at the lower end of performance and another child's standard score of 18 represented another outlier at the upper end of performance. The average value of standard scores obtained was 9.2 with a standard deviation of 3.3. Of all children tested, there were 29 who produced standard scores below 7. Of those, six produced standard scores less than 4, placing their results in the disordered category. As a result, if a binaural integration deficit were dependent upon a disordered standard score, its incidence in this group of children would be 2%. If the incidence of a deficit included children with either a disordered or a borderline standard score, the incidence would increase to 23%. Since all but the 40 children who volunteered for research participation had been referred for clinical diagnosis and were reportedly having difficulties with listening and learning, this incidence may have underestimated the incidence of binaural integration deficits in these children.***

Figure 2.  Boxplot of Competing Words subtest standard scores.

Binaural integration deficits based on interaural asymmetry

To address performance on the basis of interaural asymmetry, each child's percent correct performance in the left ear was subtracted from the percent correct performance in the right ear. The final result demonstrated either a right-ear advantage (REA) if the value was positive, or a left-ear advantage (LEA) if the value was negative.

The majority of the children who were formally assessed for handedness were right-handed, but there were three left-handed children and three children who demonstrated mixed handedness. Of those for whom handedness was assessed, all of the left-handed children and six of the right-handed children performed better in their left ear than in their right ear, thereby producing a LEA. Another sixteen children for whom no handedness information was available also produced a LEA during the Competing Words Subtest. In total, 20% of all children tested performed better in their left ear and thereby produced a negative value for interaural asymmetry. To remove the effects of laterality which could be related to handedness or attentional effects, interaural asymmetry was defined as the difference in percent correct between the two ears regardless of which ear was stronger. In this way, the value for interaural asymmetry was always positive.

Descriptive measures for the right and left ears during the Competing Words Subtest are shown in Table 1. A Kolmogorov-Smirnov test with Lilliefors significance correction showed that none of the percent correct results for the two ears were normally distributed. Again, this was not unexpected since many of the children had performed poorly on the test. A lack of normality for percent correct performance during dichotic listening tests can also occur when performance is at very high levels due to ceiling effects (Neijinhuis et al, 2002). Box plots of individual ear results and measures of interaural asymmetry obtained from the children are displayed in Figure 3. As shown in Figure 3, the lack of normality for the right ear is likely due to both ceiling effects and the presence of several outlier scores at the lower end of performance. Of the five outlier scores for the right ear, four of them were the result of poorer performance in the right ear than in the left ear, resulting in an overall LEA. Similarly, a lack of normality for interaural asymmetry results is likely due to the presence of several outlier results at the upper range of values, representing a very large difference in performance between the two ears. There were six outlier scores for interaural asymmetry ranging from 57% to 67%. There were no outliers for left ear performance because the scores covered an extremely wide range, from 0 to 97, with a mean value of 57.6 and a median score of 60.

Figure 3.  Boxplots of Competing Words subtest scores for right ear, left ear and interaural asymmetry.

Children were identified with a possible binaural integration deficit if interaural asymmetry on the Competing Words Subtest was equal to or exceeded the conservative values mentioned previously of 40% for children ages 5 and 6, 30% for children ages 7 and 8, 20% for children ages 9 to 10, and 10% for children ages 11 to 12. There were 53 children identified on the basis of large interaural asymmetries on the Competing Words Subtest. Of those 53 children, 19 of them had been previously identified on the basis of a low standard score. This meant that a total of 63 children were identified as having a binaural integration deficit by means of both standard score alone (n = 10), standard score plus interaural asymmetry (n = 19) and interaural asymmetry alone (n = 34). The use of both scoring methods more than doubled the possible incidence of a binaural integration deficit from 23% to 51%.

Referral source differences

There were 124 children involved in the study, 84 of whom had been suspected of having auditory processing difficulties and 40 of whom had volunteered for participation in auditory processing research studies. Among the 84 children who were referred for a clinical evaluation, the incidence of a binaural integration deficit by standard score alone was 32%. This relatively high incidence of children identified by standard score alone was primarily due to weak performance in both ears during the Competing Words Subtest. Because these children were suspected of having auditory processing difficulties, it was not surprising that more than half of them had difficulties with binaural integration tasks. As described previously, weak performance in both ears could have been due to a number of other factors such as language disorders or attention. After including children with weak performance in only one ear leading to a high interaural asymmetry, the incidence of binaural integration deficits in these clinically referred children increased to 54%. The remaining 40 children had volunteered for participation in research studies of auditory processing. In this group, the incidence of a binaural integration deficit by standard score alone was 2.5%. With both scoring methods, the incidence increased to 45%. This dramatic increase in the incidence of binaural integration deficits was surprising since none had been previously identified with language, listening or learning difficulties and none of their parents had previously sought a clinical evaluation of their auditory processing skills. However, when the results were discussed with the parents of the children, it was acknowledged that many of them had volunteered their children for participation because of concerns that there might have been some auditory processing difficulties that had been overlooked.

The number of children from each referral source identified under the two scoring methods is displayed in Figure 4. As shown in Figure 4, the standard scoring method was more effective among children whose listening and learning difficulties led to clinical referrals for evaluation. The interaural asymmetry method increased the number of children identified in both groups, but had a more dramatic impact on the children whose parents used the research participation opportunity for clinical information. This result highlights the importance of utilizing careful methods to screen children for their auditory processing skills prior to inclusion in research studies. If they are to be placed into either an experimental or control group, it will be necessary to validate their performance on binaural integration tasks prior to group assignment. Instructions with the SCAN-C recommend that if a child's performance on the Competing Words Subtest suggests a binaural integration deficit, the child should be tested with other dichotic listening tests to validate the presence of this type of APD. All of the children involved in research participation were additionally evaluated with the SSW and either the DDT or the RDDT and all but two of them were evaluated with the DCV. Because interaural asymmetry criteria had resulted in a higher incidence of possible binaural integration deficits, a comparison of performances across several measures was used to further validate the presence of difficulties in the children who had been evaluated with other tests. In all, this included 50 of the 124 children. Of those, 28 (or 56% of the subgroup) produced interaural asymmetry values that met or exceeded the criterion values used in this study for a possible deficit. A review of results obtained from the other dichotic listening tests validated the presence of a binaural integration deficit on at least one other test for 24 of these children, validating a dichotic listening deficit in 86% of the children who had been identified on the basis of interaural asymmetry measures.

Figure 4.  Number of children identified by standard score versus by standard score plus interaural asymmetry, separated by source of referral.

Interestingly, there were four 10-year-old children whose standard scores and interaural asymmetry values did not result in identification for a potential binaural integration deficit, but each of them performed poorly on all three of the other dichotic listening tests. There were three males, one of whom produced an interaural asymmetry of 13% and two of whom produced interaural asymmetry values of 17%. One of the males with the 17% interaural asymmetry had a standard score of 7, at the lowest level of normal for the test. It is possible that for these children, the criterion value of 20% may have been higher than needed if it was meant to identify children with potential problems. It has been demonstrated that by age 10 to 11, children's performance on dichotic listening tests approaches adult-like levels with relatively low measures for interaural asymmetry (Bellis, 2003). When children at this age appear with reports of listening and learning difficulties, it may be especially important to evaluate them with at least three dichotic listening tests to fully assess their binaural integration skills. One female performed equally in both ears, resulting in no ear advantage, and produced a standard score of 11. This child had demonstrated mixed handedness when surveyed, and her ear advantages shifted from right to left across the various tests. In her case, the pattern with reduced but inconsistent performance suggests a potential problem with attention.

Patterns of performance

There are four possible patterns of performance on the Competing Words Subtest based on standard score and interaural asymmetry. The distribution of how these four patterns occurred across the children identified with a binaural integration deficit is displayed in a pie chart in Figure 5. In the first pattern, the listeners obtained a high standard score that fell within the range of normal performance together with a low interaural asymmetry score that indicated normal relative performance between the two ears. This HiSSLoIA pattern suggests that the listener performed the task normally in both ears. Of all the children tested, 61 of them (49%) produced this pattern. If the standard score was either borderline or disordered and the interaural asymmetry score did not equal or exceed the criterion established in the study, the LoSSLoIA pattern was produced. This pattern suggested that the listener performed poorly on the task, but that the two ears performed at relatively similar levels. This is the pattern of performance that could occur for a variety of reasons since performance is similarly below normal in both ears. This pattern was produced by ten children (8%). The third pattern, HiSSHiIA, occurred when the standard score was normal, but there was a large interaural asymmetry. This pattern suggests the possibility of a binaural integration deficit type of APD since it highlights the performance difficulties in the weaker ear during competition with the stronger ear. In addition, the normal standard score is usually produced because performance in the weaker ear is well within normal limits. This auditory processing pattern was produced by 34 children (27%). A fourth and final pattern, LoSSHiIA, occurred when the overall performance for the two ears combined was below normal and there was also a high interaural asymmetry as detailed above. This result suggests the possibility of a binaural integration deficit type of APD together with the possibility of some other difficulty that resulted in reduced performance across both ears. This final pattern occurred for nineteen children (15%).

Figure 5.  Pie chart of percentage of total children who demonstrated each of the four patterns of performance on the Competing Words subtest.

The children were divided into four subgroups on the basis of their pattern of performance. Average results for the right and left ears, standard scores and measures of interaural asymmetry within each subgroup are displayed in Figure 6 through Figure 8. A multivariate analysis of variance (MANOVA) on the effects of pattern subgroup on standard scores and measures of interaural asymmetry revealed that the main effect of pattern on both types of scores was significant as expected (for standard score, F(3,123) = 70.274, p<.001; for interaural asymmetry, F(3,123) = 41.147, p<.001). Post-hoc multiple comparisons tests (Dunnett's T3) on standard score, interaural asymmetry, and individual ear scores revealed significant differences across subgroups. As shown in Figure 6, there were no significant differences in the average interaural asymmetries obtained by children between either the two subgroups with high interaural asymmetries (the first and third subgroups) or between the two subgroups with low interaural asymmetries (the second and fourth subgroups).

Figure 6.  Average values for interaural asymmetry among children within each performance pattern subgroup.

As shown in Figure 7, significant differences similarly occurred for standard score between groups that had been subdivided on the basis of either a low or high standard score (the first and second subgroup versus the third and fourth subgroup). Interestingly, however, the two subgroups that demonstrated high standard scores (the first and second subgroups in the graph) did differ significantly for standard scores as well as for interaural asymmetry. Children with high standard scores and low measures of interaural asymmetry produced an average standard score of 11.13, whereas children with high standard scores and high measures of interaural asymmetry produced an average standard score of 9.65. This statistically significant difference between the two subgroups highlights the effect that a unilateral performance decrement has on reducing the standard score even when it does not reduce it sufficiently to place it within the borderline or disordered category.

Figure 7.  Average values for standard score among children within each performance pattern subgroup.

A main effect of pattern was also revealed for right ear scores, F(3,123) = 18.129, p<.001, and for left ear scores, F(3,123) = 25.514, p<.001. Post-hoc multiple comparisons revealed that children in the two low standard score subgroups did not differ significantly for scores in their right ear nor did children in the two high standard score subgroups. As shown in Figure 8, this result suggests that right ear performance is closely tied to the child's standard score on the test. Results for the left ear were different. Children in the HiSSLoIA subgroup, whose performance was considered to be normal on the test, differed for left ear scores from children in all three other subgroups. This means that left ear performance separated the group of children with normal performance on the test from all other children who were identified with binaural integration deficits, whether the identification was made on the basis of standard score or on the basis of interaural asymmetry.

Figure 8.  Average values for right and left ears among children within each performance pattern subgroup.

Summary of results

The incidence of a potential binaural integration deficit type of APD increased from 23% based on standard scores alone to 51% based on both standard scores and measures of interaural asymmetry. Among children identified as potentially having a binaural integration deficit, slightly more than half of them (34 of 63) would not have been identified on the basis of their standard score on the Competing Words Subtest. The use of interaural asymmetry as a method for identifying children with possible binaural integration deficits significantly increased the number of children identified. When a subgroup of children (n = 28) who had been identified on the basis of interaural asymmetry were further tested with other dichotic listening tests, 86% of them had similar unilateral listening deficits on at least one other test. Since 68% of the children involved in this study had been referred for a clinical evaluation for auditory processing difficulties, an incidence of binaural integration deficits of 51% is not unexpected.

Discussion

In its original version (Keith, 1986), the full name of the SCAN was the Screening Test for Auditory Processing Disorders. For many years, audiologists and speech language pathologists used it to initially identify children who might, upon further testing, be diagnosed with auditory processing deficits. In the standardization process for the revised test, children with suspected or known auditory disorders were included in the sample to provide a better representation of processing skills across the pediatric population and the test was no longer identified as a screening measure (Keith, 2000). In spite of the revision, the test was not tested on subjects with known disorders of the central auditory nervous system, an important criterion for any test to be considered valid for diagnostic purposes. After the Dallas Consensus Conference on APD, the SCAN-C was not recommended for diagnostic use (Jerger & Musiek, 2000), leaving many clinicians unsure about its utility in their APD battery. Because the test is standardized in children and because it is familiar, commercially available, and easy to administer, it has been regarded as a useful tool for beginning the diagnostic evaluation in the audiology clinic or for screening by speech language pathologists.

A primary goal of any screening tool is to gather preliminary information about the likely presence of a deficit in a short amount of time. As described by Bellis (2003), the purpose of screening is to evaluate children who are having listening and learning difficulties in school and at home in order to determine the need for further testing. Further testing should be guided by the results from the screening process and should lead to an identification of whichever auditory processing weaknesses underlie an APD. In order to assure that most individuals with a particular deficit are identified, screening tools are designed to refer a greater number of individuals than the number who will ultimately be diagnosed with the disorder. If the Competing Words Subtest is to be useful in screening and diagnostics for binaural integration deficits, scoring techniques that increase the number of children identified as potentially having the deficit would be preferred over methods that limit the number of children identified. In this study, the standard scoring method alone failed to identify slightly more than half of the total number of children who were characterized as having a binaural integration deficit through scoring methods that included measures of interaural asymmetry.

Since the development of the first dichotic digits test (Kimura, 1961), methods that compare performance between the ears have been the preferred method for evaluating dichotic listening performance (Asbjornsen & Hugdahl, 1995; Bergman et al, 1987; Berlin et al, 1973; Blumstein et al, 1975; Bryden & Bulman-Fleming, 1994; Harper & Kraft, 1994; Hiscock et al, 2000; Hugdahl, 1995; Kershner & Morton, 1990; Repp, 1977). Performance decrements in the left ear have been reported in patients with a disruption in neural fibers of the corpus callosum (Musiek et al, 1979), in adults with auditory difficulties unexplained by hearing acuity (Jerger et al, 1991; Jerger et al, 1995; Carter et al, 2001), and in children with listening and learning difficulties (Moncrieff & Musiek, 2000; Jerger et al, 1988; Jerger et al, 1991; Jerger et al, 1999). It is not surprising, then, that performance decrements in the left ear were common among the children in this study who demonstrated weaknesses during dichotic listening whether or not they also demonstrated a large interaural asymmetry.

Because interaural asymmetry is an important measure for differentiating between listeners, it will be important to measure it consistently both within and across all dichotic listening tests. Ideally, performance by each ear should be compared to normal values from each dichotic listening test to determine if a deficit is specific only to the weaker ear during the dichotic listening task. A majority of the children identified with binaural integration deficits in this study produced normal performance in their stronger ears and reduced performance in their weaker ears. This result is suggestive of deficits specific to auditory pathways either in ascending neural structures from the side of the weaker ear or in interhemispheric pathways that transmit information from the side of the weaker ear into the language-dominant hemisphere in the cortex. Because handedness can influence the direction of the interaural asymmetry (Bryden, 1988; Dawe & Corballis, 1986; Hugdahl et al, 1990), it is helpful to assess handedness in each child evaluated for binaural integration deficits. Once handedness is known, quantifications that represent the direction of laterality are helpful when comparing performance across a number of dichotic listening tests. As shown by one child in this study, ear advantages that shift from right to left on different tests may suggest a problem with attention or motivation. Because attention and other factors may also influence performance on any single dichotic listening test, interaural asymmetry should be assessed across two tests or more to determine if results are consistent within any listener. Information about the stability of a child's performance across a number of dichotic listening tests is very important in the determination of a binaural integration deficit.

When comparing interaural asymmetry results from a single test across a large number of listeners in a group study, results will be more comparable if they are converted so that all values are positive. In this way, group averages will be accurate reflections of the degree of asymmetry between ears and will not be reduced by the presence of any negative criterion values. Values of interaural asymmetry in this study were estimates based on results from earlier studies (Keith, 2000). The identification of children based on these estimates was validated by the presence of similarly poor performance in the weaker ear on other tests of dichotic listening for which normative information was available. While this has been a first step in establishing the importance of interaural asymmetries in the determination of binaural integration deficits, a goal of future studies is to establish normal ranges of interaural asymmetry for children throughout development for all of the dichotic listening tests currently being used to assess their performance.

Comment from Robert W. Keith, Professor, University of Cincinnati, Cincinnati, OHIO (USA)

The work of the author is to be greatly admired. Moreover, the goal of her publication is laudable. However, I have concerns about parts of the article that I will summarize as follows.

My first comment has to do with a basic definition of the term ‘binaural integration.’ My overriding concern is the problem of communicating auditory processing disorders terminology because of lack of standardized definitions or imprecise use of terms. For example, the language used in describing dichotic listening tests is conflicting and, I believe, adds to confusion among persons who discuss or are learning about APD. Moncrieff, and others in the field, use the term ‘binaural integration’ when they are discussing dichotic testing. In fact, the classic definition of binaural integration is very different. For example, in 1973, Calearo and Antonelli used the terminology ‘binaural fusion/integration’ for one group of central hearing tests of incomplete message sets delivered to individual ears that are fused at the brainstem and analyzed cortically. Examples of tests that fall into that group include alternating speech tests and speech tests that present high- and low-pass filtered speech to opposite ears. The signal is then integrated at the level of the brainstem and heard as a cohesive message at cortical levels. On the other hand, central hearing tests based on binaural separation include ‘two different words delivered simultaneously with both words repeated by the subject.’ Examples of tests of binaural separation include the dichotic constant vowel tests, competing words, and competing sentences. The Staggered Spondee Word Test is another example of a test of binaural separation.

In 1970, Carhart proposed five principles that describe features of the normal flow and processing of incoming speech information within the central nervous system (CNS). Two of them apply to this discussion. They are:

• The principle of channel separation is where trains of speech are kept distinct from one another when the speech enters each ear separately. The two channels do not become confused and intertwined within the nervous system.’ This is the concept of binaural separation.

• The second principle is that ‘a single message fuses into only one message when received binaurally’. This principle is most easily demonstrated by dividing the message into two dissimilar fractions and presenting one fraction to each ear (e.g. low-pass filtered speech to one ear and high-pass filtered speech to the other, or switching a speech signal back and forth between the two ears.) This is the concept of binaural integration.

Not only is the term binaural integration used in this article contrary to what was previously described in the literature, but it is also counter-intuitive. That is, the Merriam-Webster Dictionary defines the term ‘integrate’ as ‘to form, coordinate, or blend into a functioning or unified whole.’ That definition describes perfectly what happens when incomplete message sets are presented to two ears, but does not fit with tests that require the subject to listen to words presented to each ear, separate them, recall them, and repeat them in a certain sequence. Because there is a substantial difference in opinion regarding use of the terms binaural integration (and binaural separation) some future consensus panel will have to decide about the terminology used in this manuscript.

Another concern with the manuscript is that I feel it misrepresents the scoring procedure described in the SCAN-C manual. Specifically, Moncrieff states that use of standard scores alone fails to identify children with abnormalities. The effect is to downplay the important contribution of determining whether the child's ear advantage is normal or not and adding that information when interpreting the SCAN-C results. For example, when describing the revision of SCAN to SCAN-C the author states ‘Instructions included with the SCAN-C direct the person administering the test to combine separate ear scores from the Competing Words Subtest to produce an overall raw score. The raw score is then converted to a standard score for interpretation of the results. Standard scores that fall below 7 are considered borderline if they exceed 4 and disordered if they are at 4 or below.’ What the author fails to state is that according to the test manual, a second part of the interpretation is to calculate ear advantage and prevalence scores for the competing words subtest. The manual states (page 63–64) ‘ear-advantage scores are powerful indicators of hemispheric dominance for language and neurologically based language/learning disorders. The more extreme or atypical the ear-advantage score the greater the possibility of an auditory-based disorder such as a language or learning disability.’

In addition, Moncrieff states ‘… many clinicians record only the standard score and fail to determine whether a significant difference between the two ears is present.’ There is no reference to substantiate that users of the SCAN batteries do not analyze ear differences, and so this is an undocumented opinion. It is interesting that the author never attempts to analyze the competing words data in terms of prevalence of ear advantage published in the SCAN-C manual and makes no attempt to determine whether that prevalence was normal or not. There is also no attempt made to compare abnormalities of standard scores and ear advantage as described in the SCAN-C manual to the proposed method of scoring laterality that is the subject of this paper. The author has a unique opportunity of testing construct validity, and chose to both ignore the asymmetry data available in the manual, and to talk about it further in the article. In that regard the author omitted an important part of the SCAN-C analysis in this article. It would be a more equitable comparison to obtain the standard score and asymmetry data according to instructions given in the SCAN-C manual, to determine percentage of abnormalities, and compare those data to the method proposed in the article. That analysis would allow the reader to determine whether the scoring method proposed by Moncrieff has better validity than the technique proposed in the SCAN-C manual.

As an incidental comment, the article's reference to a left-ear deficit that is linked to an APD known as an ‘integration deficit’ is followed by a reference to Keith and Engineer, 1991. Keith and Engineer never used the term ‘integration deficit’ in the referenced paper. Similarly, the article states, ‘It has therefore been recommended that any laterality score of more than 15% would suggest the possibility of a binaural integration type of APD in any child tested with the Competing Words Subtest (Keith, personal communication).’ Since I do not use the terminology ‘binaural integration type of APD’ (Gustafson and Keith, 2005) I cannot substantiate this quotation.

Regarding the data in the article, the methods section states, ‘All of the children in the clinical group were assessed according to commonly practiced guidelines in place at each of the two clinical sites and all were screened with either the SCAN or SCAN-C competing words subtest.’ As a reader I prefer to know what the guidelines of the clinical sites were, what the component tests are, and other additional information including speech-language test results and intelligence test results to include verbal and non-verbal IQ, etc. That information is important for validating whether the children suspected of having language disorders actually had that disorder.

Among other minor concerns I have with the article is that the author states that the standard test of binaural integration is the dichotic listening test and that ‘when the two ears are placed in competition, both ears should perform at roughly equivalent levels with only a small performance difference between them (Kimura, 1961).’ In fact, this statement is only true for dichotic digits. It is not true for stimuli of greater linguistic content (e.g. words, spondees, and sentences where ear differences increase for increasing linguistic content, especially for younger children.) The point is important to make because the interpretation of ear differences follows from this simplistic statement about ear differences found in tests of dichotic listening in children.

In summary, the key weakness of this study is that it fails to compare the scoring methods described in the SCAN-C manual to the scoring method proposed here. As a result, the study omits a key element in the scoring procedure and implies that there is an inherent weakness in the SCAN-C test. The premise that users of the test do not calculate ear-advantage scores is undocumented, and the implication that users would use the technique of calculating ear differences proposed here, or that this suggestion provides more discriminating data than the scoring method proposed in the SCAN-C manual is also speculative.

The idea of assessing laterality of all auditory processing test results including dichotic word testing is an important construct. However, as a reader I continue to have reservations about the lack of standard definitions for APD and the scoring procedures sometimes used for tests of APD.

Response from Deborah Moncrieff, Assistant Professor, University of Connecticut, to Comment from Robert W. Keith, Professor, University of Cincinnati

I am very grateful for the thorough comments Dr. Keith has made in response to my publication. I understand and appreciate some of his concerns and share several of them with him. He states ‘a consensus panel will have to decide about the terminology used,’ specifically with respect to the term binaural integration’, and I am sympathetic to this recommendation. As a clinician, researcher and instructor in the field of auditory processing and its assessment, I agree that the terms used to describe dichotic listening tests have been confusing. He has also expressed concerns that I neglected to evaluate and compare the effectiveness of the prevalence method for determining ear advantage with the Competing Words Subtest. The definition of binaural integration and the contrast between methods for determining ear advantage are not separate matters, but in fact are interdependent and fundamental to the manuscript. Therefore I am grateful for this opportunity to clarify these points.

In 1979, Diana Deutsch published a paper on the binaural integration of melodic patterns. In it, she described the work of early experimenters who examined limits on intelligibility from binaural alternation rates and dichotic listening with both nonverbal stimuli (Cherry & Taylor, 1954; Moray, 1975) and speech stimuli (Broadbent, 1954; Huggins, 1964; Treisman, 1971). Deutsch argued that as part of the process of binaural integration, perceptual separation is always necessary for listeners to process information arriving at the two ears, all of which are nonsimultaneous to some extent. This separation allows the listener to link acoustic elements that are highly similar and to inhibit the linkage of acoustic elements that are likely to be emanating from different sources. Therefore, she described binaural integration as a ‘mechanism invoked by listeners to prevent confusion when monitoring individual sound sources.’ An important part of this definition is that it does not suggest that for binaural integration to occur, the auditory information arriving at the two ears must be perfectly linked together into a single percept.

Cutting (1976) described binaural fusion as the phenomenon of combining two dichotic stimuli into a single perceived stimulus. Contrasted to Deutsch's definition of binaural integration as a process whereby portions of the two auditory stimuli may be either linked or maintained separately, this definition of binaural fusion appears to be limited to only those portions of the two auditory stimuli that are linked. This would imply that any test of binaural fusion is assessing the listener's ability to extract similarities between the stimuli presented to the two ears in order to arrive at a single auditory percept. On the other hand, a test of binaural integration may evaluate a listener's ability to compare and contrast the acoustic properties of the stimuli being simultaneously presented to the two ears in an effort to characterize which portions of the stimuli are perceived as a single percept and which are perceived as two separate percepts.

Both processes, binaural fusion and binaural integration, are thought to occur by means of a rapidly running cross-correlation function of auditory input at the level of the brainstem, but the outcome of this cross-correlation function may be what differentiates the two processes. Stimuli that are highly synchronous, acoustically homogeneous, and temporally aligned (such as some dichotic consonant-vowel pairs, the CVC pairs utilized in the Dichotic Rhyme Test, or some utterances with identical fundamental frequencies) will be maximally integrated by a law of common fate into one percept as though emanating from a single sound source. Under these conditions of total binaural integration, it could be said that binaural fusion has occurred. With other stimulus configurations (as in dichotic tests with digits, words, or sentences) the stimuli are more heterogeneous and would be poorly integrated, thereby resulting in little or no binaural fusion. With some binaural integration, the listener might perceive a hybrid of the two stimuli or may hybridize both of them. With little or no binaural integration, the listener could maintain two percepts of auditory stimuli having arrived from separate sources and potentially be able to identify each.

Henry Tobin (1985) referred to integration as the process whereby a speech signal alternating between the two ears (Cherry, 1953; Bocca & Calearo, 1963) must be integrated over time so that the listener can link percepts together appropriately to construct the whole message. This definition of integration does not presume that signals arriving at the two ears are fused together at the brainstem into a single percept, but rather highlights the role of cortical processes to link ascending auditory information arriving from the two ears into a whole message. The same year, Musiek and Pinheiro (1985) stated that a binaural integration task is any dichotic listening task that requires the listener ‘to respond to the stimuli presented to both ears.’ This definition of binaural integration also recognizes a listener's ability to preserve two percepts of auditory stimuli arriving from separate sources and to maintain those percepts into the cortex for linguistic processing. This definition of binaural integration, as any dichotic task with instructions to report the stimuli heard in both ears, has been utilized since in Chermak and Musiek (1997) and Bellis (2003).

What all of these researchers identified is that dichotic stimuli may fuse to a lesser or greater degree, depending primarily upon stimulus characteristics. In general, they described binaural integration as a rapid process occurring at the level of the auditory brainstem that results in sending some information to the cortex as linked (as though emanating from a single source) and other information unlinked (as though emanating from different sources). In this model, the auditory mechanism results in varying degrees of cross-correlation that yield a range of integration values, from perfect correlation as in entirely fused stimuli, to no correlation as in entirely unfused stimuli, and to any value in between.

Deutsch also raised important concerns about attention during dichotic listening when she suggested that signal detection is independent of a listener's attention. This was contrary to Repp's assertions (1977) that the primary difference between fused and unfused consonant-vowel syllables was the degree to which selective attention had been invoked. He noted that perfectly fused syllables would be heard as originating in the middle of the head and voluntary efforts to pay attention would not affect a listener's responses. The simultaneity and similarity of the acoustic properties for fused syllables would result in almost complete integration at the level of the brainstem, prior to any effects of selective attention. Alternatively, any portion of syllables that had not been fully integrated would be subject to attentional effects further upstream in the ascending auditory system. This would suggest that the effect of selective attention during dichotic listening tests is inversely related to the degree of binaural integration (or fusion) that occurs between the two stimuli. A recent study demonstrated that fewer attentional effects occurred during dichotic listening with dichotic rhymes compared to dichotic CVs, presumably because more of the dichotic rhymes had been integrated at the level of the brainstem (Shinn et al, 2005).

The situation with respect to dichotic listening gets more complicated when the task is modified to provide some control for attention. In the free recall mode, which is standard for a binaural integration task, the listener is told to listen for all stimuli and to repeat whatever is heard in both ears each time. There is no instruction to monitor one ear over the other and the listener is free to attend to the stimuli however he or she feels is appropriate. In a binaural separation task (Musiek & Pinheiro, 1985) the listener is instructed to focus entirely on one ear and ignore the other ear, thereby lowering the cognitive demands of the overall task. A third type of administration instructs the listener to repeat all stimuli in both ears (binaural integration), but to repeat the word in the right ear first in the first half of the test, and to repeat the word in the left ear first during the second half of the test. Because the listener must repeat both stimuli, it is not a separation task per se, but it instead involves a pre-cueing toward one ear at a time for each half of the test (Jerger, personal communication).

The SCAN-C Competing Words Subtest is a binaural integration test because the listener is instructed to repeat both words, but it involves pre-cueing the listener first to repeat the word in the right ear first and then later to repeat the word in the left ear first. Pre-cueing may add some elements of binaural separation to this test because the listener may be biased toward the ear that has been pre-cued. Domitz and Schow (2000) did a factor analysis of several auditory processing tests and reported the following: Compared to a dichotic digits test that loaded onto a factor for binaural integration only and the SCAN-C Competing Sentences that loaded onto a factor for binaural separation only, the Competing Words Subtest loaded onto both of the factors for binaural integration and binaural separation. One of the primary goals in the present manuscript was to review and reevaluate results from the Competing Words Subtest solely on the basis of total right-ear score and total left-ear score in order to isolate the effects of binaural integration for the children tested in this study. In the standard binaural integration format whereby the listener is directed to repeat both stimuli each time, the ear advantage obtained across the entire test is regarded to be an important measure of overall performance.

Dr Keith raised the concern that by reporting only on the measures of ear advantage obtained across the entire test, the manuscript failed to report whether subjects in this study had obtained prevalence scores for ear advantage that would have similarly identified them with a possible disorder. A main problem with the prevalence score recommended in the instructions for the SCAN-C is that it reviews the listener's ear advantage during one listening condition at a time. It is well established that many listeners are able to reverse an ear-advantage score from a REA to a LEA during a directed left listening condition (Asbjornsen & Hugdahl, 1995). Despite this reversal while attending to the left ear, normal listeners maintain a right-ear advantage when scores for the entire test are used to establish laterality. A biased ear advantage during one ‘pre-cued direction’ listening condition is typically counterbalanced in the other pre-cued condition, thereby resulting in a normally small difference between the two ears overall. As a result, any identification based on a single listening condition could potentially misidentify children whose overall interaural asymmetry was relatively small. Another important consideration for using the measure of interaural asymmetry obtained for the entire test is that all other dichotic listening tests use that method for scoring and interpreting the results.

In response to Dr Keith's concerns, I have reviewed the prevalence scores for all of the children involved in this study for whom the data are available. Prevalence scores were not obtained during the assessment process for many of the children who had been evaluated clinically. It was apparent when reviewing the files that clinicians were not evaluating prevalence as a routine part of administering the SCAN-C Competing Words Subtest. Prevalence data was available for all 52 of the children evaluated at the research lab, regardless of the reason for their referral, and was available for an additional four clinically evaluated children whose files were reviewed retrospectively.

Among the 56 children for whom prevalence data was available, 43 of them would have been identified as potentially having a problem with the Competing Words Subtest on the basis of having obtained an ear advantage in one or both of the listening conditions that had occurred 15% or less of the time during standardization of the test. This would have meant that 77% of the children tested with the Competing Words Subtest would have been suspected of having a problem with auditory processing on this measure, a result that was consistent with the manuscript's hypothesis that prevalence methods may tend to overidentify children with difficulties when they may not be present. When the possibility of a binaural integration deficit was based upon the presence of a larger-than-normal ear advantage across both listening conditions of the test as outlined in the manuscript, the number of potentially disordered children dropped to 23 out of 56, or 41%. As detailed in the manuscript, however, this was a significantly higher number of identified children than if the identification had been made only on standard score.

For all of the 20 children who would have been identified by the prevalence method but not by the interaural asymmetry method, other dichotic listening test measures were reviewed to determine if any of the children had demonstrated abnormal performance across other dichotic listening tests. If they had, this would have supported the merits of using the prevalence data to increase identification for children whose interaural asymmetry may have been within a more normal range. Only one of the 20 children had demonstrated abnormal performance on a double dichotic digits test and on the Staggered Spondaic Words Test in the left competing condition. This one child was 10 years of age and had demonstrated an interaural asymmetry on the Competing Words Subtest of 13%, the highest interaural asymmetry measured for the subtest in this group of 20 children.

One of the most challenging aspects of this manuscript was the determination of a cut-off score for normal interaural asymmetry within each age group of children. More data is definitely needed to establish how interaural asymmetry develops as children mature and to what extent a higher measure of interaural asymmetry signals a problem with the binaural integration of dichotically presented material. Tests that are widely used and highly familiar to clinicians in a large number of settings like the SCAN-C Competing Words Subtest could be used to help us produce more data on these important measures. This manuscript was written to encourage continued use of this test, utilizing this simplified scoring method to document interaural asymmetry and advance our understanding of binaural integration in school-age children.

The author thanks Jennifer Peacock Au.D. Jennifer BaiRossi and Katherine Ruffett Au.D. for their tireless efforts and enormous enthusiasm while working on this project. Preliminary results were reported at the American Academy of Audiology convention, Philadelphia, PA, April 2000. The author also wishes to thank the anonymous reviewers whose recommendations were very helpful in the development of this manuscript.