The Effects of Probe-Tone Frequency on the Acoustic-Reflex Growth Function John J. Lutolf * Honor O’Malley ** Shlomo Silman† Abstract Acoustic-reflex growth functions (ARGFs) were obtained from 20 normal-hearing listeners. Contralateral acoustic reflexes (ARs) were elicited with pure tones of 2000 Hz. The magnitude of changes in static compliant susceptance (BA) and conductance (GA) were monitored with probe-tone frequencies of 226 Hz, 678 Hz and 1000 Hz. ARGFs were obtained with six combinations of probetone frequency/admittance component: 226 Hz BA, 226 Hz GA, 678 Hz BA, 678 Hz GA, 1000 Hz BA, and 1000 Hz GA. Peak conductance (GA) and susceptance (BA) ARs were largest within the 678 Hz GA and 1000 Hz BA ARGFs, respectively. Among high-frequency probe tones, the patterns of AR growth were larger and less variable for the 678 Hz GA ARGF and the 1000 Hz BA ARGF as determined by the magnitude of their linear (b1) and quadratic (b2) polynomial coefficients and the value of their squared correlation coefficients (R2). Key Words: Acoustic reflex, acoustic-reflex growth function, probe-tone frequency, acoustic compliant susceptance, acoustic conductance, acoustic admittance, acoustic compliant reactance, acoustic resistance, acoustic impedance. Abbreviations: AR = acoustic reflex, ART = acoustic-reflex threshold, ARGF = acoustic-reflex growth function, BA = acoustic compliant susceptance, GA = acoustic conductance, YA = acoustic admittance, -XA = acoustic compliant reactance, RA = acoustic resistance, ZA = acoustic impedance, MANOVA = multiple analysis of variance, b1 = linear polynomial coefficient, b2 = quadratic polynomial coefficient, b3 = cubic polynomial coefficient, SEB = standard error of estimate of the regression coefficient, R2 = the squared multiple correlation, dB SL re: ART = dB sensation level re: the acoustic-reflex threshold. Sumario: Se obtuvieron funciones de crecimiento del reflejo acústico (ARGFs) de 20 sujetos con audición normal. Se generaron reflejos acústicos contralaterales (ARs) con tonos puros a 2000 Hz. La magnitud de los cambios en la compliancia de susceptancia estática (BA) y en la conductancia (GA) se monitorizaron con sondas de prueba a frecuencias de 226 Hz, 678 Hz y 1000 Hz. Los ARGF se obtuvieron con seis combinaciones del componente constituido por la frecuencia de la sonda de prueba y la admitancia: 226 Hz BA, 226 Hz GA, 678 Hz BA, 678 Hz GA, 1000 Hz BA y 1000 Hz GA. Los AR obtenidos a conductancia (GA) y susceptancia (BA) pico fueron los mayores dentro de las ARGF *St. John’s University, Jamaica, N.Y. 11439 and Audiology and Speech Pathology Department, Veterans Affairs New Jersey Health Care System, East Orange, New Jersey 07018 **Department of Biobehavioral Sciences, Speech-Language Pathology and Audiology Program, Teachers College, Columbia University, N.Y. 10027 † Department of Speech Communication Arts & Sciences, Brooklyn College, CUNY, Brooklyn, N.Y. 11210 Reprint Requests: John J. Lutolf, Department of Speech, Communication Sciences and Theatre, St. John’s University, 8000 Utopia Parkway, Jamaica, N.Y. 11439. 109 Journal of the American Academy of Audiology/Volume 14, Number 2, 2003 a 678 Hz GA y 1000 Hz BA, respectivamente. Entre los tonos de prueba de alta frecuencia, los patrones de crecimiento del AR fueron mayores y menos variables para los ARGF a 678 Hz GA y los ARGF a 1000 Hz BA, determinados por la magnitud de sus coeficientes de polinomio lineal (b1) y de polinomio cuadrático, y por el valor de sus coeficientes de correlación al cuadrado (R2). Palabras Clave: Reflejo acústico, función de crecimiento del reflejo acústico, frecuencia de la sonda de prueba, susceptancia de compliancia acústica, conductancia acústica, admitancia acústica, reactancia de compliancia acústica, resistencia acústica, impedancia acústica. Abreviaturas: AR = reflejo Acústico; ART = umbral del reflejo acústico; ARGF = función de crecimiento del reflejo acústico; BA = susceptancia de compliancia acústica; GA = conductancia acústica; YA = admitancia acústica; -XA= reactancia de compliancia acústica; RA = resistencia acústica; ZA = impedancia acústica; MANOVA = Análisis de variancia múltiple; b1 = coeficiente polinomial lineal; b2 = coeficiente polinomial cuadrático; b3 = coeficiente polinomial cúbico; SEB = error estándar del estimado del coeficiente de regresión; R2 = correlación múltiple al cuadrado; dB SPL re: ART = nivel de sensación en dB con relación al umbral del reflejo acústico. H igh-frequency probe tones of 678 Hz and 1000 Hz are now available on a number of immittance meters used in many clinics, among them, the Grason Stadler model GSI-33 Version 2 Middle Ear Analyzer and the, now discontinued, Virtual model 310 Immitance Meter which is still being used by many clinicians. While three studies have addressed the use of a probe tone of 660 Hz for measuring the ARGF (Creten, Vanpeperstraete, Van Camp & Doclo, 1976; Wilson & McBride, 1978; and Sprague Wiley & Block, 1981), the use of a probe-tone frequency of 1000 Hz to monitor the generation of ARGFs has not been investigated. The present study investigated the peak magnitude and growth characteristics of conductance ARGFs and compliant susceptance ARGFs, obtained with a probe tone of 1000 Hz, and compared them with compara- Table 1 Mean Peak Acoustic-Reflex Magnitude Probe Tone Frequency 226 Hz (int.- peak) Admittance BA -.183 # + (@ 21.4) GA -.131 # + (@ 19.7) (SD) (% mean) (SD/ int.- peak) (.054) (29.5%) (6.992) (.056) (42.7%) (6.165) 678 Hz (int.- peak) .346 # + * (@ 12.1) -.717 # * ^ (@ 18.2) (SD) (% mean) (SD/ int.-peak) (.298) (86.1 %) (6.504) (.365) (50.9%) (7.281) 1000 Hz (int.- peak) 1.106 + * ^ (@ 16.8) .418 # + * ^ (@ 14.0) (SD) (% Mean) (.515) (46.5%) (SD/ int.-peak) (8.089) * Sig. diff. from both 226 Hz BA and 226 Hz GA ? = < 0.05 # Sig.diff. from 1000 Hz BA ? = < 0.05 + Sig diff. from 678 Hz GA ? = < 0.05 ^ Sig. diff. from 678 Hz BA ? = < 0.05 (.509) (121%) (7.650) Note. The acoustic-reflex activating stimulus = 2000 Hz. Values = mean change in admittance in acoustic millimhos (mmho) at mean dB SL re: ART. N = 20. BA = compliant susceptance. GA = conductance. SD = standard deviation. int.- peak = intensity in dB SL re: ART at which the peak magnitude of the acoustic reflex was obtained. % mean = SD/mean X 100. SD/int.peak = standard deviation for the int.-peak value. 110 The Effects of Probe-Tone Frequency on the Acoustic-Reflex Growth Function/Lutolf et al ble measures obtained with probe-tone frequencies of 226 Hz and 678 Hz. The study sought to contribute to the development of a normative base for ARGFs obtained with a probe tone of 1000 Hz. It also sought to expand upon the findings of prior investigators relative to the characteristics of ARs obtained with a probe-tone of 660 Hz. Defining the characteristics of the ARGF for high-frequency probe tones in normals, may prove of clinical value because the characteristics of the ARGF obtained with probe tones of 678 Hz and 1000 Hz may be uniquely altered in the presence of pathological conditions (Sprague, et al., 1981). In addition, they may be more sensitive to pathological conditions than their 226 Hz counterpart. METHOD Subjects The subjects for this study were twenty young males with normal hearing. They ranged in age from 21 to 41 years (mean = 27.8 yrs). In cases where data from only 16 subjects were analyzed, due to limitations imposed by the repeated measures design, the mean age was 28.6 yrs. Air conduction and bone conduction thresholds were < 15 dB HL at octave frequencies of 250 Hz through 8000 Hz, except at 4000 Hz where they were < 20 dB HL. All air conduction and bone conduction thresholds were within ± 10 dB HL. Tympanometry and peak-pressure function were within normal limits. Instrumentation Acoustic-Reflex Activation System Contralateral acoustic-reflex growth functions were elicited with 2000 Hz puretones. These activators were generated with a Tektronix model 501 Function Generator with a rise/fall time of 25 ms, a duration of 1 s, and an inter-stimulus interval of 4 s. A stimulus train of 23 presentations was used to generate each ARGF. The intensity of the reflex-activating stimulus was presented in an ascending format from 74 dB SPL to 118 dB SPL in increments of 2 dB SPL. In addition, four silent presentation periods, each with a duration of 4.9 s, were incorporated in the stimulus train to facilitate recording of the AR. These silent periods occurred before the sixth, eleventh, sixteenth, and twenty- first stimulus presentations in the ARGF. The temporal characteristics and the intensity parameters of the acoustic-reflex activating stimuli as well as the silent periods in the stimulus train were controlled by logic modules from the Coulbourn model C-1 Modular Instrument System. Simultaneously with the onset of signal generation, the Middle Ear Analyzer was triggered. This input activated the reflexrecording circuit of the GSI-33 Version 2 for a period of 1000 msec in order to record the magnitude of the acoustic reflex. Four silent periods, each of 4.9 s were programmed within each trial. These silent periods occurred before the presentation of the reflex eliciting stimuli at the following intensity levels: 84 dB SPL, 94 dB SPL, 104 dB SPL, and 114 dB SPL. These silent periods were necessary because of constraints in the reflex-recording program of the Grason Stadler Model GSI 33 Version 2 Middle Ear Analyzer. They coincided with the time periods during which the acoustic-reflex recording circuit of the GSI-33 Version 2 was being reset to begin a new line on the recording screen. Acoustic-Reflex Recording System Changes in static acoustic compliant susceptance or static acoustic conductance at the plane of the tympanic membrane resulting from the AR, were monitored with probe tones of 226 Hz, 678 Hz and 1000 Hz. The acoustic-reflex recording system of the Grason Stadler model 33 Version 2 Middle Ear Analyzer was used to measure and record these changes in compliant susceptance and conductance. This recording system permitted generation of ARGFs incorporating six probe-tone frequency/admittance combinations, namely: 226 Hz BA, 226 Hz GA, 678 Hz BA, 678 Hz GA, 1000 Hz BA, and 1000 Hz GA. It also provided a display of these ARGFs in both graphic and digital formats. The order of presentation of each probe-tone/admittance combination was randomly selected to avoid order effects. Right and left ears were also counterbalanced. Graphic and Digital Responses The reflex recording circuit of the Grason Stadler GSI-33, Version 2, provided a graphic recording of the response and a numerical value equivalent to the peak mag- 111 Journal of the American Academy of Audiology/Volume 14, Number 2, 2003 Table 2. Summary of Average Trend Coefficients (Value in mmho) 226 Hz BA (SD) (SE) (N) Outliers = 3 Linear b1 -.0069 (.0024) (.0014) (N = 3) Quadratic b1 -.0164 (.0036) (.0013) (N = 7) Cubic b1 .0091 (.0078) (.0045) (N = 3) b2 .0004 (.0001) (.00006) b2 -.0027 (.0004) (.0002) b3 .00006 (.00004) (.00002) Sample Average - 226 Hz BA (N = 16) Curve fit to Sample ARGF Average of Sample Quad Coeffs. Quadratic Quadratic b1 b2 b1 b2 .0111 .0002 -.0110 .00020 SD (.0071) (.00038) SEB (.0006) (.00003) [R2 = .995 (SE = .0034)] [R2 = .897 (SD = .0886)] 226 Hz GA (SD) (SE) (N) Outliers = 2 Linear b1 -.0063 (.0032) (.0016 ) (N = 4) Quadratic b1 .0163 (.0101) (.0033) (N = 9) b2 .0005 (.0003) (.0001) Cubic b1 -.0161 (.0000) (N = 1) b2 .0014 (.0000) - b3 -.00004 (.0000) - Sample Average - 226 Hz GA (N = 16) Curve fit to Sample ARGF Average of Sample Quad Coeffs. Quadratic Quadratic b1 b2 b1 b2 -.0124 .0004 -.0124 .00034 SD (.0089) (.00034) SEB (.0010) (.00005) [R2 = .9836 (SE = .0052)] [R2 = .9071 (SD =.0657)] Note. SD = standard deviation. SE = standard error of estimate. N = number of subjects. BA = compliant susceptance. GA = conductance. b1 = linear coefficient. b2 = quadratic coefficient. SEB = standard error of estimate of the regression coefficient. R2 = squared multiple correlation coefficient. nitude of the response adjacent to the graphic representation of the acoustic reflex in its standard recording mode. The reflex recording circuit also permitted a digital read out of each acoustic reflex response. Response Criteria Two criteria were established to determine whether a response was considered an acoustic reflex. First, the minimum change in static conductance or compliant susceptance had to equal or exceed the following values when monitored with the following probe-tone frequencies: 1. 0.015 mmho for a probe-tone frequency of 226 Hz, 2. 0.055 mmho for a probe-tone frequency of 678 Hz. and 3. 0.085 mmho for a probe-tone frequency of 1000 Hz. 112 (Wilson, 1979; Grason-Stadler, Inc., 1998, p-30). Second, the magnitude of all subsequent responses must equal or exceed these minimum values. These minimum threshold values were those suggested by the manufacturer. The stipulation that all subsequent values in the ARGF meet or exceed those values was a conservative requirement imposed by the design of the study to insure against accepting a random false-positive change in admittance as a response. Calibration Prior to each test session, the reference intensity levels, the frequency of the probe tones, and the linearity and the frequency of the reflex-eliciting 2000 Hz stimuli were calibrated using a Bruel and Kjaer (B&K) type 2235 Precision Sound Level Meter, a B&K type 1625 1/3-1/1 Octave Filter, A Beckman The Effects of Probe-Tone Frequency on the Acoustic-Reflex Growth Function/Lutolf et al Table 3. Summary of Average Trend Coefficients (value in mmho) 678 Hz BA (SD) (SE) (N) Outliers = 3 Linear b1 .0322 (.0116) (.0069) (N = 3) Quadratic b1 .0778 (.0191) (.0110) (N = 3) Cubic b1 .1117 (.0366) (.0163) (N = 5) b2 -.0032 (.0007) (.0004) b2 -.0110 (.0034) (.0015) b3 .00032 (.00008) (.00003) Sample Average - 678 Hz BA (N = 16) Curve fit to Sample ARGF Average of Sample Quad Coeffs. Quadratic Quadratic b1 b2 b1 b2 .0465 -.0018 .0463 -.0018 SD (.0294) (.0014) SEB (.0049) (.0002) [R2 = .9531 (SE = .0242)] [R2 = .6796 (SD = .2327)] 678 Hz GA (SD) (SE) (N) Outliers = 2 Linear b1 -.0368 (.0277) (.0138 ) (N = 4) Quadratic b1 -.0154 (.0594) (.0266) (N = 5) b2 .0033 (.0021) (.0009) Cubic b1 -.1360 (.0361) (.0208) (N = 3) b2 .0114 (.0030) (.0017) b3 -.0003 (.0001) (.00006) Sample Average - 678 Hz GA (N = 16) Curve fit to Sample ARGF Average of Sample Quad Coeffs. Quadratic Quadratic b1 b2 b1 b2 -.0728 .0019 -.0761 .0021 SD (.0428) (.0020) SEB (.0045) (.0002) [R2 = .9928 (SE = .0224)] [R2 = .8794 (SD =.1452)] Note. SD = standard deviation. SE = standard error of estimate. N = number of subjects. BA = compliant susceptance. GA = conductance. b1 = linear coefficient. b2 = quadratic coefficient. SEB = standard error of estimate of the regression coefficient. R2 = squared multiple correlation coefficient. model 6240 Frequency Counter, a Hewlett Packard model 465-A amplifier, and a B&K model ASM 2cc Coupler. At the onset of the study, the probe-tone frequencies were also evaluated for harmonic distortion. In addition, the manufacturer’s calibration cavities and the internal calibration program on the Grason Stadler GSI-33 Version 2 Middle Ear Analyzer were used to calibrate the middle ear analyzer before each test session. A Gilmont model 1200 2.0 Micro-Liter Syringe was used to verify the volume and linearity of the equivalent air-volume changes associated with millimho changes in the middle-ear static compliance during reflex contraction. Statistical Analysis The peak AR magnitude (in mmho units), and the intensity level (in dB SL re: ART) at which the AR with the largest magnitude occurred were identified from the individual ARGFs. Mean and standard deviations were calculated and the data were submitted to multiple analysis of variance (MANOVA). The individual ARGFs for each subject were submitted to an analysis of trend employing the curve-fitting program available in SPSS for Windows (M.J. Norusis (1993), p.372). Within a given probe-tone frequency/admittance condition, subjects were classified according to the linear, quadratic or cubic polynomial model that provided the best fit to the data and the largest R2 value. The individual coefficients and R2 values for each member of a polynomial subgroup, within a given probe-tone frequency/admittance condition, were then averaged and a mean and standard deviation were obtained (Table 2, 3 and 4). The ARGFs for all subjects within a 113 Journal of the American Academy of Audiology/Volume 14, Number 2, 2003 Table 4 Summary of Average Trend Coefficients (Values in mmho) 1000 Hz BA (SD) (SE) (N) Outliers = 3 Linear b1 .0549 (.0179) (.0103) (N = 3) Quadratic b1 .1611 (.0545) (.0206) (N = 7) Cubic b1 .2523 (.0887) (.0512) (N = 3) b2 -.0055 (.0018) (.0007) b2 -.0211 (.0049) (.0028) b3 .0005 (.00011) (.00006) Sample Average - 1000 Hz BA (N = 16) Curve fit to Sample ARGF Average of Sample Quad Coeffs. Quadratic Quadratic b1 b2 b1 b2 .1016 -.0028 .1077 -.0027 SD (.0943) (.0045) SEB (.0057) (.0003) [R2 = .9939 (SE = .0281)] [R2 = .8709 (SD = .1389)] 1000 Hz GA (SD) (SE) (N) Outliers = 2 Linear b1 .0241 (.0173) (.0123 ) (N = 2) Quadratic b1 .0764 (.0451) (.0184) (N = 6) b2 -.0033 (.0020) (.0008) Cubic b1 -.0314 (.0055) (.0039) (N = 2) b2 .0070 (.0002) (.000157) b3 -.0002 (.0000) (.0000) Sample Average - 1000 Hz GA (N = 16) Curve fit to Sample ARGF Average of Sample Quad Coeffs. Quadratic Quadratic b1 b2 b1 b2 .0423 -.0014 .0422 -.0013 SD (.0532) (.0027) SEB (.0049) (.0002) [R2 = .9603 (SE = .0244)] [R2 = .6978 (SD =.2503)] Note. SD = standard deviation. SE = standard error of estimate. N = number of subjects. BA = compliant susceptance. GA = conductance. b1 = linear coefficient. b2 = quadratic coefficient. SEB = standard error of estimate of the regression coefficient. R2 = squared multiple correlation coefficient. given probe-tone frequency/admittance condition were averaged and the polynomial model which best fit the average ARGF, as determined by significance level and R2 value, was identified. This was invariably a quadratic trend (Table 2, 3 and 4). Finally, as with the subgroups, the coefficients from the individual quadratic models for all 16 subjects were averaged. A polynomial model was generated which contained mean b1 and b2 coefficients and a mean R2 value as well as the standard deviations associated with those measures. The coefficients of these polynomial models were used as a measure of average slope. RESULTS Peak Acoustic Reflex Magnitude The mean peak acoustic-reflex magni- 114 tudes obtained with the six combinations of probe-tone frequency/admittance component for the twenty subjects are listed in Table I. The peak magnitudes of the ARs for 226 Hz B A and 226 Hz G A were not significantly different. Peak conductance ARs assessed with a probe tone of 678 Hz were approximately five times larger than the peak conductance ARs for 226 Hz and twice as large as peak conductance ARs monitored with a probetone frequency of 1000 Hz. Conversely, peak compliant susceptance ARs monitored with a probe tone of 1000 Hz were approximately six times larger than those monitored with a probe tone of 226 Hz and three times larger than those assessed with a probe-tone frequency of 678 Hz. In addition, they showed a rapid growth rate within 10 dB SL re: ART (Fig. 1). The Effects of Probe-Tone Frequency on the Acoustic-Reflex Growth Function/Lutolf et al Acoustic-Reflex Growth Function The mean ARGFs for all six experimental conditions were best fit with a quadratic polynomial model (Figure 1, Table 2, 3 and 4). The smaller rates of growth, as defined by the mean b1 coefficient, were associated with the 226 Hz BA and 226 GAA ARGFs. The growth of compliant susceptance (BA) ARGFs was larger than the growth of conductance (GA) ARGFS for the probe tone of 1000 Hz. The reverse was true for a probetone frequency of 678 Hz. The 678 Hz BA ARGFs and the 1000 Hz GA ARGFs were characterized by extremely high variability and have no potential for clinical applications (Table 3 and 4). The average rate of growth for 1000 Hz BA ARGFs was approximately ten times larger than that of the 226 Hz BA ARGFs .The mean deceleration in growth, as indicated by the absolute value of the mean b2 coefficient, was thirteen times larger for the 1000 Hz ARGFs than the 226 Hz BA ARGFs (Table 2 and 4). The mean slope of the conductance ARGFs monitored with a probe tone of 678 Hz was approximately six-fold larger than that for the 226 Hz GA ARGFs. The average deceleration in growth was also approximately six-times larger for the 678 Hz GA ARGFs (Table 2 and 3). DISCUSSION Acoustic Reflex Magnitude The peak 678 Hz GA AR magnitudes for eleven subjects were recorded at the highest available SPL of the reflex eliciting stimulus. Therefore, these reflect maximum AR magnitudes rather than peak AR magnitudes which may have been obtained at > 118 dB SPL. This same characteristic of the data was also noted in the other five probe-tone frequency/admittance conditions but for a smaller number of subjects in each group. Peak conductance ARs assessed with a probe tone of 678 Hz were approximately six times larger than the peak conductance ARs for 226 Hz and twice as large as peak conductance ARs monitored with a probe-tone frequency of 1000 Hz (Table 1). A probe-tone frequency of 678 Hz approaches a middle ear resonance for many subjects. As such this probe-tone frequency will be more sensitive to changes in middle-ear acoustic resist- Figure 1. Mean acoustic reflex-growth functions (ARGFs) for six probe-tone frequency/admittance conditions. Each point is based on 16 observations from 16 subjects. Note. BA =compliant susceptance. GA = conductance. dB SL = dB Sensation Level re: acoustic reflex threshold. ance/conductance which accompany the AR than will probe-tone frequencies of 226 Hz or 1000 Hz (Moller, 1961, 1964; Bennett & Weatherby 1979; Bennett, 1984). This factor may explain the larger peak conductance ARs obtained with a probe tone of 678 Hz. The peak compliant susceptance ARs monitored with a probe tone of 1000 Hz were approximately six times larger than those monitored with a probe tone of 226 Hz and three times larger than those assessed with a probe-tone frequency of 678 Hz (Table 1). These findings are in agreement with those of Bennett & Weatherby (1979) who reported median values for changes in static compliant reactance when ARs for BBN were obtained at 10 dB SL re: ART. These values were 2.7 acoustic ohms for a probetone frequency of 700 Hz and -0.1 acoustic ohms for a probe tone of 1000 Hz. These results may be explained in part by the findings of Moller (1964) who observed that the AR was accompanied by a decrease in the static impedance when monitored with a probe tones above 900 Hz. Moller (1961) also noted that the middle-ear resonance near 800 Hz was abolished by the presence of an acoustic reflex. However, the higher middle-ear resonance at 1200 Hz was unaffected by acoustic-reflex contraction. The variability associated with measures of peak AR magnitude was high. When the standard deviation was expressed as a percent of the mean (SD/mean x 100), the smallest variability was observed for the 226 Hz BA and 226 Hz GA Peak ARs (29.5% and 42.7% respectively) (Table 1). The variability for 678 Hz GA and 1000 BA peak ARs approximated 50% while that for their 678 Hz BA and 1000 Hz GA counterparts exceeded 85%. 115 Journal of the American Academy of Audiology/Volume 14, Number 2, 2003 Figure 2. (a) 678 Hz GA acoustic-reflex growth functions (ARGFs) for subjects, 1-8. Figure 3. (a) 1000 Hz BA acoustic-reflex growth functions (ARGFs) for subjects, 1-8. (b) (b) 678 Hz GA ARGFs for subjects 9-16. . GA = conductance. 1000 Hz BA ARGFs for subjects 9-16. BA = compliant susceptance. Acoustic-Reflex Growth Function GA ARGFs of fifteen of sixteen subjects (93.7%) which contained a total of 149 biphasic ARs (mean = 9.3). By comparison, biphasic ARs were noted in the 1000 Hz BA ARGFs of seven of sixteen subjects (43.7%), for which a total of 45 biphasic ARs (mean = 6.4) were observed. The largest change in static admittance within a biphasic response, whether positive or negative, was recorded as the value of the AR at a given intensity increment within the ARGF. It is therefore possible that higher incidence of biphasic ARs may have contributed to the greater variability identified in the growth patterns of the 678 Hz BA and 1000 Hz GA ARGFs. The presence of multiphasic ARs was offered as an explanation by Wilson & McBride (1978) for the restricted pattern of growth observed in their mean 660 Hz ARGF. Variability In general, large inter-subject variability seems an unavoidable feature of the ARGF (Silman, et al., 1978; Wilson, 1979, 1981; Silman & Gelfand (1981); Sprague et al., 1981). Although data from the present study suggest that the 678 Hz GA ARGF and the 1000 Hz BA ARGF were accompanied by smaller inter-subject variability than their counterpart 678 Hz Bv and 1000 Hz Gv ARGFs (Tables: 2, 3 and 4), nevertheless, the relatively better variability associated wth the 678 Hz GA and 1000 Hz BA ARGFs still exceeds that which would be permissible for clinical applications. It is interesting to note that biphasic responses were recorded in the 678 Hz BA ARGFs of fifteen subjects (93.7%) who reported a total of 118 biphasic ARs (mean = 7.8). Conversely, biphasic ARs were noted in the 678 Hz GA ARGFs of six of sixteen subjects (37.5%) for whom a total of 43 biphasic ARs were observed (mean = 7.2). Biphasic AR responses were also noted in the l000 Hz 116 Probe-Tone Frequency/Admittance Effects A small difference was noted between the slopes of the 226 Hz BA and the 226 Hz GA ARGFs. Variability was also similar, as The Effects of Probe-Tone Frequency on the Acoustic-Reflex Growth Function/Lutolf et al indicated by the measures of standard deviation and standard error (Table 2). However, interaction effects were noted between the probe-tone frequency and the admittance component when compliant susceptance ARGFs and conductance ARGFs were obtained with probe-tone frequencies of 678 Hz and 1000 Hz. The steepest slope of initial growth for compliant susceptance ARGFs and the largest subsequent deceleration in their pattern of growth was observed for ARGFs obtained with a probe-tone frequency of 1000 Hz (Table 4), (Figure 3). These relationships are similar to those observed in the peak AR magnitude and are consistent with the fact that growth in the magnitude of the compliant susceptance AR is proportional to the probe-tone frequency (Moller, 1961,1964). In contrast to the compliant susceptance ARGF, the growth of the conductance ARGF was largest when monitored with a probetone frequency of 678 Hz (Table 3), (Figure. 1). The findings regarding the growth of the conductance ARGF mirror those for peak AR magnitude measures and are compatible with prior studies showing that the middleear transmission system is conductance /resistance dominated in the probe-tone frequency region of 678 Hz (Moller, 1961, 1964; Creten, et al., 1976; Wilson & McBride, 1978; Sprague, et al., 1981)). The data suggest that a probe-tone frequency of 678 Hz is better for monitoring the growth of conductance (GA) ARs while a probe-tone frequency of 1000 Hz is preferable for monitoring the growth of the susceptance (BA) AR. In both cases, the growth characteristics are larger and the variability, as defined by measures of standard deviation and standard error is smaller for these probe-tone frequency/admittance conditions. However, the magnitude of variability associated with these measures is unacceptable for clinical applications. CONCLUSION Peak AR The 678 Hz GA peak AR were approximately five times larger than peak 226 GA AR and twice as large as the peak 1000 Hz GA AR. Conversely, the 1000 Hz BA peak AR was six times larger than the 226 Hz BA peak AR and three times larger than the 678 Hz BA AR. The 678 Hz GA and 1000 Hz BA peak ARs have greater potential for diagnostic applications. However, at present, they are far too variable for clinical use ARGF The 1000 Hz BA ARGF and the 678 Hz GA ARGF had the steepest pattern of growth (as indicated by the b1 coefficient) and the largest deceleration in growth (as indicated by the b2 coefficient). Among the higher frequencies, the 1000 Hz BA and 678 Hz GA probe-tone frequency/admittance combinations provided ARGFs with less variability. However, the degree of variability was too large to encourage clinical application of these measures at the present time. Acknowledgment: We wish to thank Drs. Jane Monroe and David Krantz of Columbia University for their assistance with statistical analysis. We also wish to thank Mr. Edward Yeung, Graduate Center City University of New York for assistance with equipment design and programming. A note of appreciation is also extended to Mr. Charles “Chuck” Comitto, Northeastern Technologies Corporation, Mr. Robert Guinta, R&R Guinta Associates, and Mr. Randy Veillette, Nicolet-Grason Stadler Corporation for their help with equipment and modifications. We also wish to express our gratitude to the reviewers for their constructive comments and valuable suggestions. The data reported are partially based upon the Ph.D. dissertation of the lead author under the sponsorship of Prof. Honor O’Malley, Columbia University and the guidance of Shlomo Silman, Ph.D. (Lutolf, 1996). The study was made possible by the generosity of Mr. Sigmond K. Schwartz through the Dr. Joanne Schwartz Memorial Scholarship Fund. REFERENCES Bennett MJ & Weatherby LA. (1979). Multiple probe frequency acoustic reflex measurements. Scandinavian Audiology, 8, 233-239. Bennett MJ. (1984). Impedance concepts relating to the acoustic reflex. In S. Silman ed. The Acoustic Reflex: Basic Principles and Clinical Applications. New York: Academic Press, (pp. 35-61). Creten WL, Vanpeperstraete PM, Van Camp KJ & Doclo JR. (1976). An experimental study on diphasic acoustic reflex patterns in normal ears. Scand Audiol 5:3-8. Grason-Stadler Inc. (1998). Grason-Stadler GSI 33, Version 2 Middle Ear Analyzer: Instruction Manual 1733-0120, Rev.5. Milford, New Hampshire. Moller AR. (1961). Network model of the middle ear. J Acoust Soc Am 33:168-176. Moller AR. (1964). The acoustic impedance in experimental 117 Journal of the American Academy of Audiology/Volume 14, Number 2, 2003 studies on the middle ear. Internat Audiol 3:1-13. Silman S, Popelka GR & Gelfand SA. (1978). Effect of sensorineural hearing loss on acoustic stapedius reflex growth functions. J Acoust Soc Am 64:1406-1410. Silman S & Gelfand SA. (1981). The relationship between magnitude of hearing loss and acoustic reflex thresholds. J. Speech Hear Disord 46:312-316. Sprague BA, Wiley TA & Block MA. (1981). Dynamics of acoustic reflex growth. Audiol 20:15-40. Wilson RH & McBride LM. (1978). Threshold and growth of the acosutic reflex. J Acoust Soc Am 63:147-154. Wilson RH. (1979). Factors influencing the acoustic immittance characteristics of the acoustic reflex. J Speech Hear Res 22:480-489. Wilson RH. (1981). The effects of aging on the magnitude of the acoustic refkex. J Speech Hear Res 24:406-413. 118