Neuropsychologia 45 (2007) 1127–1133

Note

Fine grain temporal analysis in aphasia: Evidence from

auditory gap detection
G.A. Stefanatos a,∗ , Leonard E. Braitman b , Sean Madigan c
aCognitive Neurophysiology Laboratory, Moss Rehabilitation Research Institute, Albert Einstein Medical Center,
1200 W. Tabor Road, Philadelphia, PA 19141, United States
b Office of Research and Technology Development, Albert Einstein Health Care Network, Philadelphia, PA, United States
c Department of Linguistics, University of Delaware, Newark, DE, United States

Received 17 August 2005; received in revised form 12 September 2006; accepted 19 September 2006
Available online 3 November 2006

Abstract
Auditory temporal processing was investigated in individuals with acquired aphasia using a task in which they were asked to detect brief silent
gaps inserted between noise segments modeled after formants in speech. To examine within-channel gap detection, gaps of 10, 20, 40, and 80 ms
duration were inserted between an initial segment (IS) and a trailing segment (TS) centered at the same frequency (1 kHz). In a between-channel
gap detection condition, gaps of 20, 40, 80, and 100 ms duration were inserted between an IS that differed in frequency (4 kHz) from the TS (1 kHz).
The effect of gap onset timing was examined in both conditions by systematically varying the duration of the IS by 10, 20, or 40 ms. A combined
analysis revealed that for both conditions and all gap and IS durations, individuals with aphasia produced fewer correct responses than age-matched
neurologically intact controls. Separate condition analyses revealed that when noise segments were centered at the same frequency, individuals with
aphasia demonstrated poorer accuracy in detecting 40 and 80 ms gaps relative to normal controls (p < 0.001). When gaps were inserted between noise
segments differing in frequency, on average, aphasic subjects performed less accurately at durations of 40, 80 and 100 ms (p < 0.025). Detection
in both groups decreased with smaller IS durations. The difficulties with gap detection observed in the aphasic group suggest the existence of
fundamental problems in processing the temporal form or microstructure of sounds characterized by rapidly changing onset dynamics.
© 2006 Elsevier Ltd. All rights reserved.

Keywords: Aphasia; Perception; Auditory; Temporal processing; Left hemisphere; Gap detection; Temporal resolution

1. Introduction suggesting that the left hemisphere’s superiority in phonologi-

cal processing may be based in functional asymmetries in more
Clinical and experimental studies of functional asymme- elemental aspects of auditory temporal processing that mediate
tries in the human brain have suggested that the left cerebral the analysis of both speech and comparably complex nonspeech
hemisphere is particularly adept at processing speech sounds sounds (Creese, 1999; Moore, 2000; Schwartz & Tallal, 1980).
such as stop-consonants which require the accurate perceptual Gap detection is a fundamental aspect of temporal process-
elaboration of short-term spectrotemporal variations in sound ing considered particularly relevant to this proposition because
(Dwyer, Blumstein, & Ryalls, 1982; Shankweiler & Studdert- brief silent gaps in speech serve as important cues to percep-
Kennedy, 1967; Stefanatos, Gershkoff, & Madigan, 2005; Tyler, tually distinguish voiced and unvoiced stop consonants (e.g.,
1992). Early theoretical accounts of this phenomenon postu- /ba/ vs. /pa/, /da/ vs. /ta/, /ga/ vs. /ka/) (Lisker, 1975; Lisker &
lated the existence of lateralized linguistic mechanisms spe- Abramson, 1964; Stevens, 1998). In addition, psychophysical,
cializing in phonetic analysis (Liberman, Cooper, Shankweiler, electrophysiological and functional neuroimaging studies have
& Studdert-Kennedy, 1967; Studdert-Kennedy & Shankweiler, suggested a preferential role for the left hemisphere in gap detec-
1970). Alternative explanations were subsequently marshaled tion (Brown & Nicholls, 1997; Nicholls, Gora, & Stough, 2002;
Sulakhe, Elias, & Lejbak, 2003; Vroon, Timmers, & Temelaars,
1977) involving neural networks shared with the perception
∗ Corresponding author. Tel.: +1 215 456 5962. of stop consonants (Zaehle, Wustenberg, Meyer, & Jancke,
E-mail address: Stefanag@Einstein.edu (G.A. Stefanatos). 2004).

0028-3932/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuropsychologia.2006.09.011
1128 G.A. Stefanatos et al. / Neuropsychologia 45 (2007) 1127–1133

An association between gap detection and speech processing between sounds that do not overlap in frequency composition,
has also been suggested by clinical studies showing that chil- then detection entails a temporal correlation of the relative tim-
dren identified with specific language impairment (SLI) demon- ing of the discontinuation of activity in the perceptual channels
strate problems with gap detection (Ludlow, Cudahy, Bassich, representing the pre-gap sound and the onset of activity in the
& Brown, 1983; McCroskey & Kidder, 1980). Moreover, longi- perceptual channels representing the post-gap sound (Phillips
tudinal developmental studies have shown that gap detection and Hall, 2002; Phillips & Smith, 2004). Thresholds in this
performance in infancy is predictive of subsequent levels of “between-channel” (BC) form of gap detection are often an
language comprehension and production in toddlers (Trehub & order of magnitude larger than WC thresholds (Formby et al.,
Henderson, 1996) and preschoolers (Benasich & Tallal, 2002; 1998; Neff, Jesteadt, & Brown, 1982; Phillips and Hall, 2002;
Benasich, Thomas, Choudhury, & Leppanen, 2002). Further- Phillips et al., 1997) and are influenced to a greater extent by
more, animal models have linked poor gap detection to cortical temporal factors such as gap onset timing. Gaps occurring early
neuromigrational anomalies of the type observed in individuals in an auditory stream (e.g., 5 ms), for example, are associated
with SLI (e.g., ectopias and microgyria) (Clark, Rosen, Tallal, & with detection thresholds that are two to four times higher than
Fitch, 2000a, 2000b; Peiffer, Friedman, Rosen, & Fitch, 2004) thresholds for detecting gaps in the middle or towards the end
as well as to bilateral lesions of auditory cortex (Bowen, Lin, of auditory stimuli (Eggermont, 1999; Phillips et al., 1997).
Taylor, & Ison, 2003; Kelly, Rooney, & Phillips, 1996; Syka, Phillips and colleagues (see Phillips, 1999) have argued that
Rybalko, Mazelova, & Druga, 2002). detecting gaps early in narrowband stimuli more closely resem-
Despite this convergence of evidence, studies of acquired bles the requirements of processing gap cues in speech than
language disorders in adults suggest that unilateral left hemi- the broadband stimuli typically used in traditional gap detec-
sphere lesions do not result in remarkable difficulties with gap tion paradigms. In order to assess gap detection in acquired
detection. Divenyi and Robinson (1989) observed that individ- aphasia in a way that that would provide a closer analog of
uals with acquired aphasia were not significantly different from the type of temporal processing relevant to the perception of
neurologically intact controls on a traditional gap detection task speech, we examined early gap detection using syllable-length
where brief silent gaps were inserted in the middle of 200-msec narrowband noise bursts modeled after formants in speech. We
bursts of broadband white noise. Similarly, Robin et al. (1990) compared the performance of individuals with chronic aphasia
Robin, Tranel, and Damasio (1990) observed that gap detection to age-matched neurologically normal controls using both WC
in patients with left posterior infarcts was not different from sub- and BC gap detection tasks, systematically varying gap duration
jects with posterior right hemisphere lesions. These findings are and gap onset timing.
at variance with suggestions that problems with speech process-
ing in aphasia are functionally linked to general problems with 2. Methods
auditory temporal resolution.
One possible explanation for the apparent inconsistency in 2.1. Subjects
these findings is that engagement of specialized left hemisphere
neural mechanisms may vary depending on the spectrotempo- Thirteen individuals (nine males and four females) with aphasia, aged 28–68
years (mean = 49.69, S.D. = 11.61), were recruited from the volunteer aphasia
ral properties of stimuli and specific processing requirements subject pool coordinated through the Aphasia Center at Moss Rehabilitation
(Nicholls, Schier, Stough, & Box, 1999; Sulakhe et al., 2003). Research Institute. According to the results of the Western Aphasia Battery
Psychophysical experiments have shown that the spectrotempo- (Kertesz, 1982), five had Broca’s aphasia, seven were classified with Anomic
ral attributes of the sounds in which gaps are embedded can aphasia, and one had Wernicke’s aphasia that resolved to pure word deafness.
determine the specific perceptual operations engaged in gap Severity spanned the mild to moderate range with Aphasia Quotients ranging
between 45.8 and 92.2 (mean = 74.3). All were native English speakers who
discrimination and influence detection thresholds (Fitzgibbons, demonstrated evidence of aphasia of more than 3 months duration. The eti-
1983; Florentine & Buus, 1984; Snell, Ison, & Frisina, 1994). ological basis of the aphasia was limited to single events such as stroke or
Based on such findings, two forms of gap detection have been hemorrhage in the distribution of the left middle cerebral artery. Twelve sub-
distinguished, each mediated by different neural mechanisms jects demonstrated structural damage including but not limited to the superior
(Phillips, 1999; Phillips, Taylor, Hall, Carr, & Mossop, 1997). temporal lobe and in eight cases the lesion encroached on Heschl’s gyrus. The
remaining subject had a lesion involving the left insula extending into subcorti-
When silent gaps are inserted in a homogeneous sound (e.g., cal white matter. Additional inclusion criteria were: (1) normal range nonverbal
200 ms broadband noise burst), detection is contingent upon IQ (>80); (2) high school education or equivalency; (3) normal range hear-
the capacity of the nervous system to resolve an interruption ing (audiometric pure tone thresholds less than 25 dB at 0.5, 1 and 2 kHz and
in the neural excitation of the same perceptual channels acti- less than 30 dB at 5 kHz); (4) premorbid right-hand preference ascertained by
vated before and after the silent period (Eddins & Green, 1995; the Edinburgh Handedness Inventory (operationalized as a laterality quotient
>33). Exclusion criteria were: (1) current or premorbid history of a neuro-
Plomp, 1964; Zeng, Kong, Michalewski, & Starr, 2005). In this logical condition with chronic cognitive sequelae aside from those implicit in
“within-channel ” (WC) form of gap detection, thresholds are the event resulting in aphasia (i.e. learning disability, developmental language
extremely brief, in the order of a few milliseconds for broad- disorder, pervasive developmental disorder, encephalitis, multiple sclerosis, sig-
band stimuli (Eddins, Hall, & Grose, 1992; Formby, Gerber, nificant closed head trauma); (2) history of drug or alcohol abuse or dependence
Sherlock, & Magder, 1998; Moore, Peters, & Glasberg, 1993; within the 6 months preceding the study; (3) evidence of an inability to com-
ply with experimental task demands. As a comparison group, we also studied
Plomp, 1964), and somewhat higher for narrowband sounds 13 neurologically intact right-handed controls matched for age (mean = 48.92,
(Grose, 1991; Grose, Eddins, & Hall, 1989; Schneider, Pichora- S.D. = 10.26), gender (nine males and four females) and socioeconomic
Fuller, Kowalchuk, & Lamb, 1994). By contrast, if gaps occur status.
 G.A. Stefanatos et al. / Neuropsychologia 45 (2007) 1127–1133 1129

presentation of the gap and “no gap” comparator within each pair varied across
successive trials. Subjects were asked to ascertain if the stimuli comprising
the pair were the “same” or “different” and to indicate this by pointing to an
appropriately designated box on a response card. The ‘same/different’ response
was chosen over a two-alternative forced-choice procedure to avert the need for
subjects to remember the temporal order or interval in which stimuli were pre-
sented. Prior to each experimental run, instructions were delivered to subjects
via a multimedia PowerPoint presentation, which supplemented audioverbal
and written directions with pictures. Additionally, to ensure that subjects’ ade-
quately comprehended the requirements of the task, and to reduce bias and
anchor same/different responses (Rousseau, Meyer, & O’Mahony, 1998), they
were exposed to the standard and signal stimuli and administered a series of
practice trials with feedback.

2.4. Statistical analysis

For purposes of statistical analysis, the number of correct detections was

Fig. 1. Schematic diagram depicting stimulus characteristics in the within- calculated for each gap and IS duration for left and right ears of each subject.
channel and between-channel gap detection paradigm. In within-channel gap Exploratory analyses revealed no statistical difference between ears in overall
detection (upper), the center frequency of sound segments before and after the accuracy so we subsequently collapsed across ears. Because the range of gap
silent gap were both 1 kHz. In between-channel gap detection (lower), the cen- durations differed in the WC and BC conditions, we performed separate anal-
ter frequency of the sound segment before the gap was 4 kHz while the trailing yses for each condition (312 observations in each analysis). The independent
segment was 1 kHz. In the within-channel condition, initial segments of 10, 20, variables for the analyses in the WC condition were group (aphasia, control),
and 40 ms duration were paired with gap durations of 10, 20, 40 and 80 ms. In IS duration (10, 20 and 40 ms) and gap duration (10, 20, 40, and 80 ms). The
the between-channel condition, these same initial segments were paired with BC analyses took the same form except different levels of gap duration (20, 40,
gap durations of 20, 40, 80 and 100 ms. On each trial, these stimuli were paired 80 and 100 ms) were used. In both conditions, the frequency distribution of the
with an equivalent stimulus with a 0 ms gap duration. percentage of correct responses (accuracy) deviated substantially from a normal
distribution (both p < 0.0005 by the Shapiro–Wilk test of normality). Rather than
use an analysis of variance (ANOVA), which assumes normality, we analyzed
2.2. Instrumentation and apparatus the data using multivariable binomial logistic regression. This is an analytic
method specifically designed for situations where the dependent variable is a
BC and WC gap detection were examined by inserting silent gaps in band- percentage (Collett, 1991; McCullagh & Nelder, 1989) and does not assume a
limited (200 Hz bandwidth) noise segments modeled after formants in speech. normal distribution. Since the number of correct detections out of 8 trials was a
Stimuli were digitally synthesized with a phonetics software package (Boersma, repeated measure (over each combination of gap and IS duration), we adjusted
2003) using a 20 kHz sampling rate and 16-bit precision. Each stimulus consisted for the correlation that arises from repeated observations on the same subject
of a brief initial segment (IS) and a longer (250 ms) trailing segment (TS) sepa- (analogous to adjustments made in repeated measures ANOVA).
rated by a silent gap. The rise/fall time for each segment was 0.5 ms. In the WC We wanted to compare the effects of condition (WC and BC) directly and
gap detection task, the IS and TS were both centered at 1 kHz and separated by investigate possible interactions by condition. To do this, we also performed a
gaps of 10, 20, 40, or 80 ms duration. In the BC condition, IS segments were statistically more powerful combined binomial logistic regression analysis (468
centered at a different frequency (4 kHz) from TS segments (1 kHz). BC gap observations) considering gap durations common to both conditions (20, 40 and
detection is more difficult (Phillips et al., 1997) and pilot data suggested that 80 ms). All analyses assessed the effects of group, gap duration, IS duration and
longer gap durations were necessary to avoid potential floor effects. We there- all possible interactions. The statistical analyses were completed using Stata
fore paired gaps of 20, 40, 80, and 100 ms duration with the same IS and TS version 8 (StataCorp., 2003).
durations used in the WC condition. The effect of gap onset timing was exam-
ined in both conditions by systematically varying the IS duration by 10, 20, or
40 ms, holding the TS constant at 250 ms. 3. Results
Each gap stimulus was coupled with a corresponding “no gap” standard stim-
ulus comprised of the same IS and TS separated by a 0-ms gap. Since the 0.5 ms Unadjusted average percent correct scores for each group
rise/fall times was retained, a discontinuity was present in the amplitude enve- (aphasia vs. control), gap duration, and IS duration are depicted
lope of the 0-ms gap stimulus at the point of conjunction of the segments. This
was preferred as a “no gap” comparator over a continuous band-limited noise
in Fig. 2a for the WC condition and Fig. 2b for the BC condition.
as it reduces the likelihood that stimuli containing gaps could be discriminated
simply by detecting the spectral cues occurring at the onset and offset of each seg- 3.1. WC gap detection
ment (Phillips et al., 1997). The inter-stimulus interval was 800 ms. A schematic
representation of the stimuli used in each condition is depicted in Fig. 1.
Stimuli were presented monaurally to subjects through a pair of Sennheiser We analyzed all 312 percentages (13 subjects × 4 gap dura-
HD 240 stereo headphones at 60 dB above an estimated threshold based on tions × 3 IS durations × 2 groups) in the WC condition in a
audiometric screening. WC and BC gap detection conditions were examined in multivariable binomial logistic regression analysis, which con-
separate blocks of trials. Each block of 96 trials assessed 12 different combina- sidered the effect of each variable simultaneously controlling
tions of gap duration (4) and IS duration (3). Each combination was assessed
for all other variables. This analysis revealed a main effect of
eight times (four in each ear). Trials were presented in pseudorandom order
separated by a 2000-ms inter-trial interval. IS duration (p < 0.001) indicating that mean accuracy increased
with IS duration for all gap durations. The only statistically
significant interaction was between gap duration and group
2.3. Task
(p = 0.02). This indicated that the relationship between gap dura-
On each trial, subjects were presented with a pair of stimuli comprised of tion and accuracy differed in the two groups. Within the aphasic
a stimulus containing a gap and a standard “no gap” comparator. The order of group and within the normal controls, mean accuracy increased
1130 G.A. Stefanatos et al. / Neuropsychologia 45 (2007) 1127–1133

Fig. 2. (a) Unadjusted group-averages of percentage of correct responses in the within-channel gap detection task. For each IS duration, the percentage of correct
responses increases with gap duration in subjects with aphasia and in age-matched normal controls. A small increase in percent of correct responses is apparent with
increases in IS duration, evident in comparison of the panels from left to right. (b) Group averages of percentage of correct responses in the between-channel gap
detection task. Again, the percentage of correct responses increases with gap duration and IS duration in both aphasic group and age-matched normal controls.

with gap duration (both p < 0.001). In order to compare accuracy duration (100 ms) and the BC condition did not include the low-
between the aphasic and control groups, we performed separate est gap duration (10 ms). For both conditions and all gap and IS
analyses for gap durations of 10, 20, 40 and 80 ms. The aphasic durations, the aphasic group averaged fewer correct responses
and control groups had similar difficulty detecting short duration than the controls (p < 0.001). On average, subjects in both groups
gaps (for 10 and 20 ms, both p > 0.13). At larger gap durations had fewer correct responses in BC than in WC discrimination
(40 and 80 ms), the aphasic group performed less accurately for all gap durations in this comparison (p < 0.001). In addition,
than controls (both p < 0.001). None of the other two-way or the main effect of IS duration was significant (p < 0.001), indi-
three-way interactions was statistically significant (all p > 0.6). cating that both groups demonstrated better performance with
increasing IS duration.
3.2. BC gap detection In this combined WC–BC analysis, the interaction between
gap and group was not significant (p = 0.14), indicating that con-
Multivariable binomial logistic regression analysis of the BC dition did not modify the effect of any of the other variables. This
condition revealed a statistically significant (p = 0.001) interac- contrasts with the statistically significant interactions between
tion between gap duration and group, again indicating that the gap and group in both separate condition analyses. Furthermore,
relationship between gap duration and accuracy differed in the none of the other two-, three- or four-way interactions were sig-
two groups. Separate analyses in the two groups revealed that nificant (all p > 0.17).
mean accuracy increased with gap duration in both the aphasic
and control groups (both p < 0.001). In order to directly compare 4. Discussion
accuracy in the aphasic and control groups in this condition,
we performed separate analyses for gap durations of 20, 40, These results indicate that individuals with acquired aphasia
80 and 100 ms. The aphasic group performed worse than the secondary to left hemisphere lesions showed poorer process-
normal control group at all gap durations (40, 80 and 100 ms) ing of short silent gaps in syllable-length narrowband noise
(p < 0.025), except for the shortest gap duration (20 ms, p > 0.7), bursts modeled after formants in speech. For both conditions
where both groups performed poorly. Again, mean accuracy and all gap and IS durations, individuals with aphasia averaged
increased with IS duration (p = 0.001). fewer correct responses than age-matched neurologically intact
controls. The manifestation of these difficulties varied with the
3.3. Combined WC and BC gap detection spectrotemporal properties of the sounds delimiting the gaps.
One of the key influences affecting performance on both WC
The foregoing analyses examined WC and BC gap detection and BC tasks was gap duration. As gap duration increased, the
separately because the range of gap durations differed between mean number of correct responses increased in both the control
conditions. In order to compare accuracy in the two conditions, and aphasic subjects. However, when conditions were analyzed
we performed analyses considering only gap durations com- separately, there were statistically significant group by gap inter-
mon to both (20, 40 and 80 ms). This statistically more powerful actions in both WC and BC gap detection. Analysis of the
combined binomial logistic regression analysis was based on interaction in the WC gap detection condition revealed that both
468 observations. However, it used only a subset of the gap groups demonstrated poor accuracy at the short gap durations
durations because the WC condition did not use the highest gap (10 and 20 ms) whereas individuals with aphasia demonstrated
 G.A. Stefanatos et al. / Neuropsychologia 45 (2007) 1127–1133 1131

poorer accuracy than controls in detecting longer gap durations have suggested that BC gap detection requires time-stamping
(40 and 80 ms). Similarly, in the BC condition, group differences and temporal correlation of activity in different perceptual chan-
were evident at larger gap durations (40, 80 and 100 ms). It there- nels. This process is thought to take place centrally and require
fore appeared that the groups differed in the rate of increase of greater allocation of attentive and perceptual resources. Indi-
correct responses with increasing gap duration. This difference, viduals with aphasia often demonstrate limitations in allocation
evident in the multivariable analyses, was also reflected in a of attention and perceptual resources (Erickson, Goldinger, &
rightward shift of the psychometric functions derived for the LaPointe, 1996; Korda & Douglas, 1997; Murray, 2002) and
aphasic group relative to those of the control group (see Fig. 2). this may have contributed to the their difficulties on the BC
This shift suggests that, compared to controls, the individuals gap detection task. Alternative explanations for their impaired
with aphasia in this study required longer duration gaps to reach performance may include problems with duration processing
threshold levels of performance (75% correct) and had difficul- (Ilvonen et al., 2001; Nichelli, Clark, Hollnagel, & Grafman,
ties in processing suprathreshold gaps. 1995), time tagging (Efron, 1963), or the utilization of tempo-
The results on the WC gap detection task differ from Divenyi ral information (Halbig, Yves von Cramon, Schmid, Gall, &
and Robinson (1989) who observed that the performance in Friederici, 2002; Lalonde & Hannequin, 1999).
subjects with aphasia was comparable to normal controls. Dif- Studies of the neurological correlates of auditory temporal
ferences in the spectrotemporal characteristics of the noise in processing in humans suggest that neural processes involved in
which gaps were embedded may account for this disparity. WC and BC gap detection are elaborated cortically in posterior
Broadband noise, as used by Divenyi and Robinson (1989), stim- temporal cortex including primary auditory cortex (Heinrich,
ulates a wide range of frequency channels, and consequently Alain, & Schneider, 2004) and the planum temporale (Zaehle
the detection of gaps is facilitated by the presence of temporal et al., 2004). However, Efron, Yund, Nichols, and Crandall
discontinuities across multiple channels. By contrast, gaps in (1985) described modest increases in gap detection thresholds
narrowband noise activate fewer channels and accordingly gaps in patients who had undergone temporal lobectomy sparing pri-
must be longer in duration in order to be detected with reason- mary auditory cortex and planum temporale. The effect was
able certainty. This may be particularly true when stimuli are limited to the contralesional ear at or near threshold levels of
below 5–6 kHz, as we used in this study, because of the longer performance using gap durations of 40 and 50 ms. The absence
integration times associated with this frequency range. As a con- of an ear effect in the present study may be due to several fac-
sequence, we observed generally higher gap detection thresholds tors including smaller sample size, heterogeneity of lesion size
(>20 ms) in both groups and the aphasic group appeared to have and localization, as well as a number of procedural differences.
particular difficulty with these discriminations. However, our findings are consistent with Efron et al. in sug-
The statistically significant effect of IS duration suggests that gesting that deficits in gap detection may occur from lesions
the onset timing of gaps in the temporal stream also influenced that compromise regions other than primary auditory cortex and
performance. This was evident in separate analyses of the WC the planum temporale.
and BC conditions, as well as in the combined condition analysis. In conclusion, these findings are consistent with evidence
The basis of the difficulties demonstrated by the aphasic group suggesting that left hemisphere lesions impede perceptual
is not entirely clear but may possibly reflect anomalous cod- discriminations requiring the precise analysis of informa-
ing of brief initial sound segments or of rapidly changing onset tion that unfolds or changes rapidly over time (25–50 ms)
dynamics. This may also relate to observations that individuals (Boemio, Fromm, Braun, & Poeppel, 2005; Liebenthal, Binder,
with left hemisphere lesions demonstrate difficulty resolving Spitzer, Possing, & Medler, 2005; Liegeois-Chauvel, de Graaf,
rapidly presented pairs of clicks (Albert & Bear, 1974; Lackner Laguitton, & Chauvel, 1999; Stefanatos et al., 2005; Zatorre &
& Teuber, 1973) which are extremely brief (typically <1 ms). Belin, 2001). They also provide important information about
By contrast, they have little difficulty on tasks where gaps are hemispheric asymmetries in gap detection. If the left and right
placed in the middle of broadband stimuli lasting a couple of hemisphere were equipotential for rapid temporal analysis, then
hundred milliseconds (Divenyi & Robinson, 1989; Robin et al., the intact right hemisphere in our subjects with aphasia might be
1990). expected to mediate normal gap detection performance. To the
Visual comparison of the psychometric functions suggests extent that this was not the case, our findings provide support
differences between groups may be larger at the higher gap dura- for the notion of left hemispheric involvement in mediating early
tions in the BC than in the WC condition (Fig. 2a and b). How- gap detection. However, strong claims regarding the superiority
ever, the multivariable combined condition analysis revealed no of the left hemisphere in gap detection cannot be made with-
evidence of such a statistical difference in accuracy for those out a direct comparison to performance in subjects with right
gap durations common to both conditions. That analysis, by hemisphere lesions.
necessity, excluded the largest and smallest gap durations. Such
a difference may exist at those higher gap durations, but the Acknowledgements
design of this study precludes addressing this question.
The marked decrement in performance in the BC condition The research in this paper was supported by a grant from
is consistent with suggestions that different perceptual opera- the Albert Einstein Society. Correspondences can be addressed
tions mediate BC and WC gap detection (Phillips, 1999; Phillips to Gerry A. Stefanatos, D. Phil., Director, Cognitive Neuro-
and Smith, 2004; Phillips et al., 1997). Phillips and colleagues physiology Laboratory, Moss Rehabilitation Research Institute,
1132 G.A. Stefanatos et al. / Neuropsychologia 45 (2007) 1127–1133

Albert Einstein Medical Center, 1200 W. Tabor Rd., Philadel- Grose, J. H. (1991). Gap detection in multiple narrow bands of noise as a function
phia, PA 19141, USA. Electronic mail can be sent to Ste- of spectral configuration. Journal of the Acoustical Society of America, 90(6),
fanaG@Einstein.edu. 3061–3068.
Grose, J. H., Eddins, D. A., & Hall, J. W., 3rd. (1989). Gap detection as a function
of stimulus bandwidth with fixed high-frequency cutoff in normal-hearing
References and hearing-impaired listeners. Journal of the Acoustical Society of America,
86(5), 1747–1755.
Albert, M. L., & Bear, D. (1974). Time to understand. A case study of word Halbig, T. D., Yves von Cramon, D., Schmid, U. D., Gall, C., & Friederici, A.
deafness with reference to the role of time in auditory comprehension. Brain, D. (2002). Processing of temporal duration information in working memory
97(2), 373–384. after frontodorsal tumour excisions. Brain and Cognition, 50(2), 282–303.
Benasich, A. A., & Tallal, P. (2002). Infant discrimination of rapid auditory cues Heinrich, A., Alain, C., & Schneider, B. A. (2004). Within- and between-channel
predicts later language impairment. Behavioural Brain Research, 136(1), gap detection in the human auditory cortex. Neuroreport, 15(13), 2051–2056.
31–49. Ilvonen, T. M., Kujala, T., Tervaniemi, M., Salonen, O., Naatanen, R., & Pekko-
Benasich, A. A., Thomas, J. J., Choudhury, N., & Leppanen, P. H. (2002). The nen, E. (2001). The processing of sound duration after left hemisphere stroke:
importance of rapid auditory processing abilities to early language develop- Event-related potential and behavioral evidence. Psychophysiology, 38(4),
ment: Evidence from converging methodologies. Developmental Psychobi- 622–628.
ology, 40(3), 278–292. Kelly, J. B., Rooney, B. J., & Phillips, D. P. (1996). Effects of bilateral auditory
Boemio, A., Fromm, S., Braun, A., & Poeppel, D. (2005). Hierarchical and cortical lesions on gap-detection thresholds in the ferret (Mustela putorius).
asymmetric temporal sensitivity in human auditory cortices. Nature Neuro- Behavioral Neuroscience, 110(3), 542–550.
science, 8(3), 389–395. Kertesz, A. (1982). The western aphasia battery. New York: Grune & Stratton.
Boersma, P. (2003). PRAAT 4. 1. Amsterdam: University of Amsterdam. Korda, R. J., & Douglas, J. M. (1997). Attention deficits in stroke patients
Bowen, G. P., Lin, D., Taylor, M. K., & Ison, J. R. (2003). Auditory cortex with aphasia. Journal of Clinical and Experimental Neuropsychology, 19(4),
lesions in the rat impair both temporal acuity and noise increment thresh- 525–542.
olds, revealing a common neural substrate. Cerebral Cortex, 13(8), 815– Lackner, J. R., & Teuber, H. L. (1973). Alterations in auditory fusion thresholds
822. after cerebral injury in man. Neuropsychologia, 11(4), 409–415.
Brown, S., & Nicholls, M. E. (1997). Hemispheric asymmetries for the temporal Lalonde, R., & Hannequin, D. (1999). The neurobiological basis of time esti-
resolution of brief auditory stimuli. Perception and Psychophysics, 59(3), mation and temporal order. Reviews in the Neurosciences, 10(2), 151–173.
442–447. Liberman, A. M., Cooper, F. S., Shankweiler, D. P., & Studdert-Kennedy,
Clark, M. G., Rosen, G. D., Tallal, P., & Fitch, R. H. (2000a). Impaired processing M. (1967). Perception of the speech code. Psychological Review, 74(6),
of complex auditory stimuli in rats with induced cerebrocortical microgyria: 431–461.
An animal model of developmental language disabilities. Journal of Cogni- Liebenthal, E., Binder, J. R., Spitzer, S. M., Possing, E. T., & Medler, D. A.
tive Neuroscience, 12(5), 828–839. (2005). Neural substrates of phonemic perception. Cerebral Cortex, 15(10),
Clark, M. G., Rosen, G. D., Tallal, P., & Fitch, R. H. (2000b). Impaired two- 1621–1631.
tone processing at rapid rates in male rats with induced microgyria. Brain Liegeois-Chauvel, C., de Graaf, J. B., Laguitton, V., & Chauvel, P. (1999). Spe-
Research, 871(1), 94–97. cialization of left auditory cortex for speech perception in man depends on
Collett, D. (1991). Modelling binary data. London: Chapman & Hall. temporal coding. Cerebral Cortex, 9(5), 484–496.
Creese, I. (1999). Rate processing constraints may underlie developmental lan- Lisker, L. (1975). Is it VOT or a first-formant transition detector? Journal of the
guage impairments and also hemispheric specialization for speech. Brain Acoustical Society of America, 57, 1547–1551.
Research Bulletin, 50(5–6), 431–432. Lisker, L., & Abramson, A. S. (1964). A cross-language study of voicing in
Divenyi, P. L., & Robinson, A. J. (1989). Nonlinguistic auditory capabilities in initial stops: Acoustical measurements. Word, 20, 384–422.
aphasia. Brain and Language, 37(2), 290–326. Ludlow, C. L., Cudahy, E. A., Bassich, C., & Brown, G. L. (1983). Auditory
Dwyer, J., Blumstein, S. E., & Ryalls, J. (1982). The role of duration and rapid processing skills of hyperactive, language-impaired, and reading disabled
temporal processing on the lateral perception of consonants and vowels. boys. In E. Z. Lasky, & J. Katz (Eds.), Central auditory processing disorders
Brain and Language, 17(2), 272–286. (pp. 163–184). Baltimore: University Park Press.
Eddins, D. A., & Green, D. M. (1995). Temporal integration and temporal res- McCroskey, R. L., & Kidder, H. C. (1980). Auditory fusion among learning
olution. In B. C. J. Moore (Ed.), Hearing. San Diego: Academic. disabled, reading disabled, and normal children. Journal of Learning Dis-
Eddins, D. A., Hall, J. W., 3rd, & Grose, J. H. (1992). The detection of temporal abilities, 13, 69–76.
gaps as a function of frequency region and absolute noise bandwidth. Journal McCullagh, P., & Nelder, J. A. (1989). (2nd ed.). Generalized linear models:
of the Acoustical Society of America, 91(2), 1069–1077. vol. 37, London: Taylor & Francis CRC Press.
Efron, R. (1963). Temporal Perception. Aphasia and déjà vu. Brain, 86, 403–424. Moore, B. C., Peters, R. W., & Glasberg, B. R. (1993). Detection of temporal
Efron, R., Yund, E. W., Nichols, D., & Crandall, P. H. (1985). An ear asymmetry gaps in sinusoids: Effects of frequency and level. Journal of the Acoustical
for gap detection following anterior temporal lobectomy. Neuropsychologia, Society of America, 93(3), 1563–1570.
23(1), 43–50. Moore, D. R. (2000). Auditory neuroscience: Is speech special? Current Biology,
Eggermont, J. J. (1999). Neural correlates of gap detection in three auditory 10(10), R362–R364.
cortical fields in the cat. Journal of Neurophysiology, 81(5), 2570–2581. Murray, L. L. (2002). Attention deficits in aphasia: Presence, nature, assessment,
Erickson, R. J., Goldinger, S. D., & LaPointe, L. L. (1996). Auditory vigilance and treatment. Seminars in Speech and Language, 23(2), 107–116.
in aphasic individuals: Detecting nonlinguistic stimuli with full or divided Neff, D. L., Jesteadt, W., & Brown, E. L. (1982). The relation between gap dis-
attention. Brain and Cognition, 30(2), 244–253. crimination and auditory stream segregation. Perception and Psychophysics,
Fitzgibbons, P. J. (1983). Temporal gap detection in noise as a function of fre- 31, 493–501.
quency, bandwidth, and level. Journal of the Acoustical Society of America, Nichelli, P., Clark, K., Hollnagel, C., & Grafman, J. (1995). Duration process-
74(1), 67–72. ing after frontal lobe lesions. In J. Grafman, K. J. Holyoak, & F. Boller
Florentine, M., & Buus, S. (1984). Temporal gap detection in sensorineural and (Eds.), Annals of the NY academy of science (pp. 183–190). New York, NY:
simulated hearing impairments. Journal of Speech and Hearing Research, Academy of Sciences.
27(3), 449–455. Nicholls, M. E., Gora, J., & Stough, C. K. (2002). Hemispheric asymmetries
Formby, C., Gerber, M. J., Sherlock, L. P., & Magder, L. S. (1998). Evidence for visual and auditory temporal processing: An evoked potential study.
for an across-frequency, between-channel process in asymptotic monaural International Journal of Psychophysiology, 44(1), 37–55.
temporal gap detection. Journal of the Acoustical Society of America, 103(6), Nicholls, M. E., Schier, M., Stough, C. K., & Box, A. (1999). Psychophysical
3554–3560. and electrophysiologic support for a left hemisphere temporal processing
 G.A. Stefanatos et al. / Neuropsychologia 45 (2007) 1127–1133 1133

advantage. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, Snell, K. B., Ison, J. R., & Frisina, D. R. (1994). The effects of signal frequency
12(1), 11–16. and absolute bandwidth on gap detection in noise. Journal of the Acoustical
Peiffer, A. M., Friedman, J. T., Rosen, G. D., & Fitch, R. H. (2004). Impaired gap Society of America, 96(3), 1458–1464.
detection in juvenile microgyric rats. Brain research. Developmental Brain StataCorp. (2003). Stata, Release 8. College Station, TX: Stata Corporation.
Research, 152(2), 93–98. Stefanatos, G. A., Gershkoff, A., & Madigan, S. (2005). On Pure Word Deafness,
Phillips, D. P. (1999). Auditory gap detection, perceptual channels, and tem- temporal windows and the left hemisphere. Journal of the International
poral resolution in speech perception. Journal of the American Academy of Neuropsychological Society, 11, 456–470.
Audiology, 10(6), 343–354. Stevens, K. (1998). Acoustic phonetics. Cambridge: MIT Press.
Phillips, D. P., & Hall, S. E. (2002). Auditory temporal gap detection for noise Studdert-Kennedy, M., & Shankweiler, D. (1970). Hemispheric specialization
markers with partially overlapping and non-overlapping spectra. Hearing for speech perception. Journal of the Acoustical Society of America, 48(2),
Research, 174(1–2), 133–141. 579–594.
Phillips, D. P., & Smith, J. C. (2004). Correlations among within-channel and Sulakhe, N., Elias, L. J., & Lejbak, L. (2003). Hemispheric asymmetries for
between-channel auditory gap-detection thresholds in normal listeners. Per- gap detection depend on noise type. Brain and Cognition, 53(2), 372–
ception, 33(3), 371–378. 375.
Phillips, D. P., Taylor, T. L., Hall, S. E., Carr, M. M., & Mossop, J. E. (1997). Syka, J., Rybalko, N., Mazelova, J., & Druga, R. (2002). Gap detection thresh-
Detection of silent intervals between noises activating different perceptual old in the rat before and after auditory cortex ablation. Hearing Research,
channels: Some properties of “central” auditory gap detection. Journal of 172(1–2), 151–159.
the Acoustical Society of America, 101(6), 3694–3705. Trehub, S. E., & Henderson, J. L. (1996). Temporal resolution in infancy and
Plomp, R. (1964). Rate of decay of auditory sensation. Journal of the Acoustical subsequent language development. Journal of Speech and Hearing Research,
Society of America, 36, 277–282. 39(6), 1315–1320.
Robin, D. A., Tranel, D., & Damasio, H. (1990). Auditory perception of temporal Tyler, L. (1992). Spoken language comprehension: An experimental approach
and spectral events in patients with focal left and right cerebral lesions. Brain to disordered and normal processing. Cambridge: MIT Press.
and Language, 39(4), 539–555. Vroon, P. A., Timmers, H., & Temelaars, S. (1977). On the hemispheric
Rousseau, B., Meyer, A., & O’Mahony, M. (1998). Power and sensitivity of the representation of time. In Paper presented at the attention and perfor-
same-different test: Comparison with triangle and duo-trio methods. Journal mance VI: Proceedings of the international symposium on attention and
of Sensory Studies, 13, 149–173. performance.
Schneider, B. A., Pichora-Fuller, M. K., Kowalchuk, D., & Lamb, M. (1994). Zaehle, T., Wustenberg, T., Meyer, M., & Jancke, L. (2004). Evidence for
Gap detection and the precedence effect in young and old adults. Journal of rapid auditory perception as the foundation of speech processing: A sparse
the Acoustical Society of America, 95(2), 980–991. temporal sampling fMRI study. European Journal of Neuroscience, 20(9),
Schwartz, J., & Tallal, P. (1980). Rate of acoustic change may under- 2447–2456.
lie hemispheric specialization for speech perception. Science, 207(4437), Zatorre, R. J., & Belin, P. (2001). Spectral and temporal processing in human
1380–1381. auditory cortex. Cerebral Cortex, 11(10), 946–953.
Shankweiler, D., & Studdert-Kennedy, M. (1967). Identification of consonants Zeng, F. G., Kong, Y. Y., Michalewski, H. J., & Starr, A. (2005). Perceptual con-
and vowels presented to left and right ears. Quarterly Journal of Experimen- sequences of disrupted auditory nerve activity. Journal of Neurophysiology,
tal Psychology, 19(1), 59–63. 93(6), 3050–3063.