BINAURAL FUSION TEST

Matzker (1959) developed a binaural rest in which Ger­man two-syllable PB words were filtered through a low-pass band (500—800 Hz) in one ear and a high-pass band (1815-2500 Hz) in the other ear. The high-pass band was presented simultaneously with the iow-pass band. Normal subjects are able to fuse the high-pass band and low-pass band and recognize the word. That is, they integrate the information at the two ears (summate binaurally). When only one of the bands (high- or low-pass) is presented to an ear, normal subjects cannot recognize the word. This test is considered a special case of dichotic speech tests since dif­ferent but complementary stimuli are presented simulta­neously to the two ears. Matzker's procedure involved first dichotic presentation, then diotic presentation (which did not require binaural summation or fusion), followed again by dichotic presentation. In normal-hearing subjects, per­formance in the dichotic mode is essentially equivalent to performance in the diotic mode. Matzker hypothesized that patients with brainstem pathology, particularly lower brainstem pathology, would perform poorly on the dicho­tic mode of the binaural-fusion test; that is, they would be unable to fuse the high-and low-pass bands of informa­tion.

Linden (1964) developed a Swedish version of the Matz­ker binaural-fusion test based on Swedish spondaic words. His low-pass band was from 560 to 715 Hz and his high-pass band was from 1800 to 2200 Hz. Linden hypothesized that patients with various central pathologies would obtain binaural-fusion test scores in the dichotic mode lower than those when high-and low-pass bands were presented simul­taneously to one ear. The results did not bear out his hy­pothesis. Linden concluded that his modification of the binaural-fusion test was not useful for the detection of central auditory pathology.

In Ivey's (1969) version of the binaural-fusion test, the low-pass band extended from 500 to 700 Hz and the high-pass band extended from 1900 to 2100 Hz. The rejection rate of the filter was 36 dB per octave. Two lists of 20 children's spondaic words were presented with the low-pass band at 25 dB SL relative to the 500-Hz pure-tone threshold and the high-pass band at 25 dB SL relative to the 2000-Hz threshold. The first list was given with the low-pass band to one ear and the high-pass band to the other ear. With the second list, the ears receiving the high- and low-pass bands were reversed. Each item was assigned a value of 5%. Arbi­trarily, the binaural-fusion score is attributed to the ear in which the low-pass band is presented. Ivey (1969) adminis­tered the binaural fusion test to a group of normal-hearing listeners. The mean score for list 1 was higher than that for list 2 (93.8% and 86.0%, respectively.

Willeford (1977) modified the Ivey (1969) version of the binaural-fusion test, using the presentation of 30 rather than 25 dB SL (higher sensation levels if low scores are obtained at 30 dB SL). Willeford (1977), as noted by Lukas and Gendeur-Lukas (1985), recommended that a 10% cor­rection factor be added to score:; obtained for list. 2. Mon­aural recognition is first obtained for each frequency band for each ear. Then the binaural-fusion recognition scores are obtained with one list presetted such that the right ear receives the low-frequency band and the other list presented such that the left ear receives the low-frequency band. Willeford (1977) reported that the mean binaural-fusion score for his group of 20 normal adults was 89% with a range of 75-100%. According to Lynn and Gilroy (1975), normal binaural-fusion scores usually surpass the highest monaural single-band score by at least 50%.

Lynn and Gilroy (1972) reported that performance on the binaural-fusion test was abnormal in some of their patients with temporai-lobe tumors. Nevertheless, in some cases there was evidence of secondary brainstem compres­sion. In a later study on 4 patients with posterior temporal-lobe tumors, Lynn and Gilroy (1975) reported that abnormal binaural-fusion test scores were obtained in 1 patient and questionable bmaurai-fusion test scores were obtained in another patient. Lynn and Gilroy (1975) also observed that the binaural-fusion test results were ab­normal in only 1 of their 6 patients with antero-inferior temporal-lobe tumors, 0 of their 5 patients with superfi­cial parietal-lobe tumors, and 1 of their 8 patients with deep parietal-lobe tumors (with a questionable score in another patient). These results suggest that the binaural-fusion test is not sensitive to temporal-lobe or parietal-lobe tumors.

In the Smith and Resnick (1972) version of the binaural-fusion test, monosyllables (CNC) were employed. The low-frequency band was 360—890 Hz and the high-frequency band was 1750-2220 Hz. The test was presented under three conditions; (a) dichotic with tbe low-pass band to one ear and the high-pass band to the opposite ear, (b) diotic with high- and low-pass bands presented binaurally, and (c) the same dichotic condition as in (a) except that the ears receiving the low- and high-pass bands were reversed. Smith and Resnick (1972} found that performance on the binaural-fusion test was similar across the three conditions in subjects with normal hearing-thresh old levels, bilateral sensorineural hearing impairment, and temporal-lobe pa­thology. In all 4 of their subjects with brainstem pathology, performance in at least one of the two dichotic conditions was reduced compared with that in the diotic condition. Smith and Resnick's (1972) findings suggested that the binaural-fusion test could be used to detect brainstem pa­thology. Palva and Jokinen (1975) reported that poor binaural-fusion test scores (with good scores when the low-and high-pass bands were presented monaurally) occurred frequently in patients with brainstem lesions of vascular or traumatic etiology. On the other hand, Lynn and Gilroy (1977) did not observe consistent findings or patterns on the binaural-fusion test in patients with brainstem lesions. Also, Musiek and Geurkink (1982), who used the Willeford (1977) version of the binaural-fusion test, reported that only 30% of their 10 patients with brainstem lesions dem­onstrated abnormal binaural-fusion test scores. These later reports suggest that the binaural-fusion test has a low sensi­tivity to brainstem lesions, contrary to what was originally theorized. Large-sample studies on the binaural-fusion test are required before this test can be used clinically to detect brainstem dysfunction.

RAPIDLY ALTERNATING SPEECH PERCEPTION TEST (RASP)

Bocca and Calearo (1963) developed a swinging speech test following the method of Cherry and Taylor (1954). In the swinging speech test, the message alternates periodically between ears so that each ea.: receives half of the message. Recognition of the entire message occurs only if the infor­mation to the two ears is fused or resynthesized. Figure 13 illustrates the poor recognition in each ear alone with com­plete recognition when the two ears receive the comple­mentary segments.

Bocca and Calearo (1963) found that performance on the rapidly alternating speech perception (RASP) test was unaf­fected by pathology of the auditory cortex. Since RASP performance was affecred by some cases of diffuse cerebral pathology and many cases of brainstem pathology, they suggested that the RASP test was dependent upon the integ­rity of the centra] auditory nervous system at the level of the brainstem.

Lynn and Gilroy (1975) developed their own English language version of the Bocca and Calearo swinging speech test. The stimuli alternated every 300 ms. Normative data collected in their laboratory revealed monaural scores rang­ing from 0-10% and binaural scores from 95-100%. In a later study based on 18 normal adults, Lynn and Gilroy (1977) obtained mean monaural scores of 7.2 and 3.9% for the right and left ears, respectively, with standard devia­tions of 9.0 and 5.1% for the right and left ears, respec­tively; in the binaural condition, the mean score was 97.2% and the standard deviation was 7.4%. The Willeford (1977) version of the RASP test differed from the Lynn and Gilroy (1975) version essentially with respect to the stimu­lus items. The presentation leve] is 30 dB SL relative to the pure-tone average. Some investigators have recommended

BINAURAL RESYNTHESIS RAPIDLY ALTERNATING SPEECH PERCEPTION

RE:     I'          EE   YO         GH     AF          L     CH

LE:       LL   S              U   Ri     T        TER      UN

BIN:     I'LL   SEE   YOU   RIGHT   AFTER   LUNCH

figure 13 Binaural resynthesis sample from the Rapidly Alternating Speech Perception Test.

the use of 40 dB SL relative to the SRT (Tobin, 1985) or 50 dB SL relative to the SRT or PTA (Musiek & Geurkink, 1982; Pinheiro, 1978). The procedure involves first pre­senting the sentences monaurally to each ear (with a score for each ear) and then presenting the sentences in the al­ternating mode with the two ears receiving the comple­mentary segments. Twenty sentences are presented, each sentence having an assigned value of 5%. The sentences used in the RASP are shown in Table VI. In the alternating mode, 10 sentences are presented so that the right ear re­ceives the first segment and 10 sentences are presented so that the left ear receives the first segment. Willeford (1977) reported that the mean scores for the RASP in a group of 20 normal adults was 99% with a range of 90-100%.

Lynn and Gilroy (1977) reported that the RASP mean score was 100% in their group of 11 patients with right temporal-lobe tumors, approximately 80% for their group of 11 patients with left temporal-lobe tumors, approxi­mately 92% for their group of 14 patients with right pa­rietal-lobe tumors, approximately 92% for their group of 10 patients with left parietal-lobe tumors, approximately 38% for their group of 6 patients with low pons cerebello-pontine-angle lesions, and approximately 84% for their group of 9 patients with upper brainstem lesions. These results show that the poorest RASP scores were obtained in patients with low brainstem lesions. Musiek and Geurkink (1982) reported abnormal scores on the RASP test in only 50% of their 10 patients with brainstem lesions. Antonelli et al. (1987) obtained findings similar to those of Musiek

Table VI   Sentences from the Rapidly Alternating Speech Perception Test"

SENTENCE

BEGINS______

TRACK

L     I. The children came home lace from school.

L     2. He likes to play and splash in the water.

R    3. She is mad because the television was broken.

R    4.  It rained very much for hours.

R    5.  The wind almost blew the roof off.

L    6. What time did you get in yesterday?

L     7. The school room was very large.

R    8.  I took a picture of her with my camera.

L     9. What is your favorite television program?

R  10. Where have you been all day?

L   II. The telephone rang for a real long time.

L 12. How did you put that together again?

L   13. They went to the park to play games.

L   14.  Let the dog in at the back door.

R  15. Answer the telephone as soon as it rings.

R  16.  Please turn the radio up louder.

L   17. Are you going to the office this morning?

L  18. That window is very dirt}'.

R  19.  His father is on the police force.

R 20.  I had bacon and eggs for breakfast.

and Geurkink (1982). Antonelli et al. (1987) reported that approximately 56% of their 16 cases with intra-axial brainscem lesions had abnormal alternating speech test scores whereas only approximately 6% of their 16 cases with temporal-lobe lesions had abnormal alternating speech scores.

SPEECH WITH ALTERNATE MASKING INDEX (SWAMI)

  Given by Jerger

  Check for brainstem pathology

  Present speech signal binaurally and masking noise is alternated between 2 ears.

  Here also binaural SIS is founding presence of noise.

  There is a noise burst in both ears alternatively so our ear will be receiving masked signal and other receiving good signal.

  Normally normal subjects will get 100% score because of binaural integration.

  In subjects with brain stem lesion results will be poorer.

  Signalis presented at 40-50dBSL

  0dB SNR or less is used.

  Prefebarly PB list or monosyllable are used.

  Jerger originally used PB words for this test.

MASKING LEVEL DIFFERENCE TEST

Binaural advantages in detection of signals in masking noise have been investigated in various clinical populations. These binaural advantages are also known as binaural release from masking, masking level differences (MLDs), or binaural unmasking. The mechanism for extraction of the signal from a background of noise is believed to be analysis of interaural phase (time) differences (Durlach &c Colburn, 3978).

Consider the case of a person who has a noise (N) and a signal (S) such as speech presented to one ear. This condi­tion, which reflects monotic presentation of the signal and noise, has been represented in the literature as SmNm. The condition in which the signal is presented to one ear but the noise is presented to both ears with a 180° interaural phase difference is represented as SmNπ. The condition in which the signal is presented to one ear but the noise is presented to both ears with a 0° interaural phase difference is repre­sented as SmNo. The condition in which the signal is pre­sented to both ears with a 0° interaural phase difference and the noise is presented to both ears with a 0^○ interaural phase difference is represented as SoKo. The condition in which the signal is presented to both ears with a 180° interaural phase difference and the noise is presented to both ears with a 0° interaural phase difference is  represented as SהּNo. The condition in which the signal is presented to both ears with a 0° interaural phase difference and the noise is presented to both ears with a 180° interaural phase difference is repre­sented as S0Nהּ. The condition in which the signal is pre­sented to both ears with a 180° interaura] phase difference and the noise is presented to both ears with a 180° phase difference is represented as SהּNהּ.

Figure 14 illustrates some of these experimental condi­tions. The S0N0 condition is hcmophasic since the noise and signals are in phase. The SoNהּ, SהּNo, and SהּNהּ are antiphasic conditions since the noises and/or the signals are out of phase.

According to Diercks and Jeffress (1962), the SmNm, S0N0, and SהּNהּ produce the poorest thresholds. The 

SoNהּ and SהּNo produce the best thresholds. This situation is illustrated by the magnitude of the smile or frown in Figure 14. In normal-hearing persons, the threshold for a signal in the SmNo or SmNהּ condition is better than that in the SmNm condition. Also, the threshold for a signal in the SoNהּ or SהּNo condition is better than that in the SoNo condition. These improvements reflect release from mask­ing. The MLD can be obtained by subtracting the threshold level for the signal in the SmNo or SmNהּ condition from the threshold level for the signal in the SmNm condition. The MLD can also be obtained by subtracting the threshold level for the signal in the SoNהּ or SהּNo condition from the threshold level for the signal in the SoNo condition.

The MLD for speech is most commonly measured by obtaining the spondee recognition threshold in the SoNo condition (reference condition) and SoNהּ or SהּNo condi­tion. The size of the MLD is defined as the difference be­tween the SRT in the bomophasic condition and that in the antiphasic (SהּNo or SoNהּ) condition. The intensity of the broad-band noise is usually presented at 80 dB SPL (Lynn, Gilroy, Taylor, & Leiser, 1981; Olsen & Noffsinger, 1976; Olsen, Noffsinger, & Carhart, 1976). Some two-channel audiometers (e.g., GSI-10) have a network that allows phase reversal of either the noise or test signal. Adaptors with the mixer and phase shift capability are commercially available for use with two-channel audiometer and can be purchased from some companies.

Olsen and Noffsinger (1976) obtained mean spondee MLDs of 7.3 dB for the SהּNo antiphasic condition and 6.9 dB for the SoNהּ antiphasic condition when the refer­ence condition was SoNo in their group of 12 normal-hearing subjects. Olsen et al. (1976) obtained a mean spondee MLD of 8.3 and 6.9 dB for the SהּNo and SoNהּ conditions, respectively, in their group of 50 normal-hearing subjects. Since only 6% obtained spondee MLDs less than or equal to 5 dB for the SהּNo condition and less than or equal to 3 dB for the SoNהּ condition, the cutoff MLD for normalcy below which the MLD was abnormally reduced was 5 dB for the former condition and 3 dB for the latter condition. Lynn et al. (1981) reported that the mean MLD in their group of normal-hearing subjects was 12.2 dB for the SהּNo antiphasic condition and 10.2 dB for the SoNהּ antiphasic condition; the standard deviation was 1.1 dB for both antiphasic conditions and the range was 10-14 dB for the SהּNo antiphasic condition and 8-12 dB for the SoNהּ antiphasic condition.

Cullen and Thompson (1974) reported that the MLD score (% Michigan CNC words correct in SoNהּ or SהּNo condition minus % Michigan CNC words correct in the SoNo condition) in 4 patients with anterior temporal-lobe lesions involving the superior gyrus fell in the range defining one standard deviation above and below the mean in normal-hearing listeners. Cullen and Thompson (1974) concluded that an intact auditory association was not es­sential for release of masking to occur. They also concluded that the release from masking phenomenon was probably mediated at subthalamic levels. Lynn et al. (1981) reported that the mean MLD detection threshold in their group of 12 patients with cerebral lesions did not differ significantly from that of the normal subjects for both SהּNo and SoNהּ antiphasic conditions. Olsen et al. (1976) reported that abnormally small spondee MI.Ds were obtained in only 5% of their 20 patients with cortical lesions for the SהּNo anti-phasic condition and 0% of these patients for the SoNהּ antiphasic condition. Olsen et al. (1976) concluded that the MLD was mediated at subcorcical levels. Noffsinger (1982) reported that the spondee MLD (SoNo - SהּNo) was ab­normally small (less than 6 dB) in only 9% of their 67 cases with temporal-lobe lesions. The results of these investiga­tions on the MLD in temporal-lobe lesions show that the MLD is generally unaffected by cortical lesions.

Olsen and Noffsinger (1976) reported that 58% of their 12 patients with multiple sclerosis and other pathologies of the midbrain and/or brainstern region had abnormlly small spondee MLDs of less than 6 dB for the SהּNo antiphasic condition and/or less than 4 dB in the SoNהּ antiphasic condition. They concluded that the MLD was often im­paired by central-nervous system pathologies associated with multiple sclerosis or other midbrain and/or brainstem lesions. Noffsinger (1982) reported that 52% of their 114 patients with brainstem lesions and normal-hearing sensitivity had spondee MLDs (SoNo minus SהּNo) that were abnormally reduced (<6 dB). Noffsinger (1982) concluded that abnormally reduced MLDs in persons with normal hearing-thresholds signaled the presence of brainstem pa­thology. Lynn et al. (1981) reported that the mean MLDs for the speech-detection threshold did not differ signifi­cantly   among   the   normal,   cerebral,   and   upper   pons/ midbrain/thalamic groups. The mean MLD was signifi­cantly reduced in the pontomedullary group compared with the normal, cerebral, and upper pons/midbrain/thaiamic groups. Lynn et al. (1981) concluded that the pontomed­ullary region of the brainstem (lower brainstem) mediates that MLD phenomenon and suggested that the region of the superior olivary nuclei was a likely site for mediation of the MLD since it was the most caudal anatomical site in the brainstem where there was integration of information from the two ears. The hypothesis of Lynn et al. (1981) was substantiated by the findings of Noffsinger, Martinez, and Schaefer (1982). Noffsinger et al. (1982) found that ab­normally small MLDs and abnormally elevated or absent acoustic-reflex    thresholds    occurred    in    patients    with brainstem lesions when abnormalities in the brainstem au­ditory-evoked potentials began with wave I, II, or III but normal MLDs and acoustic-reflex thresholds occurred in patients with brainstem lesions when abnormalities in the brainstem auditory-evoked potentials began with wave IV or V. Since waves I and II probably arise from the eighth cranial nerve, and the primary contribution to wave III is the superior olivary complex according to Britt and Rossi (1980) and Rossi and Britt (1980) or the cochlear nuclei according to Moller, Jannetta, and Moller (1981), Noff­singer et al. (1982) proposed that cochlear nuclei and/or the superior olivary complex may be responsible for the binaural interaction processes involved in the MLD phenome­non. Since abnormally reduced spondee MLDs occur in patient   with   eighth-nerve   tumors   even   when   there   is normal-hearing sensitivity and symmetrical audiometric configurations (Noffsinger, 1982; Olsen et al., 1976), the presence  of a reduced MLD in  persons with  bilateral normal-hearing thresholds and symmetrical audiometric configurations does not differentiate between eighth nerve and brainstem sites of lesion. The spondee MLD is associ­ated with a higher sensitivity than the 500-Hz MLD to brainstem pathology (Olsen & Noffsinger, 1976; Noff­singer, 1982).

Nevertheless, since the hit rate for the spon­dee MLD  appears not to be substantially higher than chance, the spondee MLD test does not appear to hold great promise as a clinical tool for the detection of brainstem dysfunction in normal-hearing persons.

Several studies have shown reduced speech MLDs in pa­tients with multiple sclerosis (Hannley, Jerger, & Rivera, 1983; Olsen & Noffsinger, 1976; Olsen et al, 1976).