- Email: [email protected]
The effect of presentation rate on the comprehension and recall of speech after anterior temporal-lobe resection
Pergamon
THE EFFECT OF PRESENTATION RATE ON THE COMPREHENSION AND RECALL OF SPEECH AFTER ANTERIOR TEMPORAL-LOBE RESECTION INGRIII JOHNSRUDE and BRENDA MILNER Department of Psychology and Department of Neurology and Neurosurgery, McGill Llniversity; and the Montreal Neurological Institute, 3801 University St, Montreal. Quebec, Canada H3A ?B4 (Rewired
I9 March
1993; rrccepted 7 July 1993)
Abstract-Twenty-two
paticnfs who had undergone unilateral anterior resection of either the left (LT) or right (RT) temporal lobe and 13 normal control (NC) subjects were tested on a lexicaldecision task and a story-memory task. Stimuli were presented aurally. and. in the latter task, at five different speech rates ranging from 125 words per minute (wpm) to 325 wpm. Memory for stories in the LT group was not abnormally sensitive to the effect of increasing rate. although it was inferior to that of the NC group at all speeds. This argues against the hypothesis that the verbal memory defcit evident after LT resection is partially attributable to an abnormally slow processing of verbal information. But LT subjects were impaired at judging the lexicality of uords and nonwords presented aurally, suggestmg that the left anterior temporal region plays a special role in the processing of speech sounds.
INTRODUCTION IT IS WELL ESTABLISHEDthat left anterior temporal-lobe resection impairs memory for verbal material. Indeed, impoverished memory for short prose passages (four or five sentences in length) remains, in our experience, the most conspicuous deficit exhibited by patients with such removals [6]. This deficit may, in part, be attributable to impaired processing of the material to be remembered: in particular, indirect evidence from several studies [ 12, 16, 191 indicates that patients with a left anterior temporal-lobe excision are impaired relative to those with right temporal-lobe excisions and normal subjects at assimilating and making use of information under time pressure. The question therefore arises as to whether, after left anterior temporal-lobe resection. patients are able to process verbal information as rapidly as normal subjects. This hypothesis was first suggested by BLAKEMOREand FALCONER [a], and was examined by FRISK and MILNER [7] in the visual modality. FRISK and MILNER [7] assessed comprehension of single words and short prose passages at three different presentation rates, using a rapid serial visual presentation (RSVP) procedure. For all groups, comprehension of stories decreased with increasing rate of presentation, and, although the left temporal-lobe group with large hippocampal removals made more errors than other groups when recognizing story content, their performance was not differentially impaired by increasing rate of presentation, relative to that of other groups. Despite these negative findings for visually presented material, the possibility remains that different results might be obtained in the auditory modality. The neural substrates of auditory and visual language processing are not identical [8, 143, and converging evidence
from several studies suggests that the left anterior temporal ncocortex plays ;I special role in the re-evocation of the sounds of recently heard words [ I3 161. Thus, excisions from this region may affect the rate at which spoken Innguagc is proccsscd. The prcscnt study-. dcsigncd to test this hypothesis, is an adaptation of FRISK and MILXR’S [7] procedure to the auditory modality. Normal English-speaking subjects find a speech rate of about 165 wpm the easiest to understand 191, and rutcs below 139 wpm and abovz 189 wpm arc judged to be “slow” and “fast”. respectively [S]. In the present study. sentences and stories wcrc prcscntcd aurally at slow ( I25 wpm) normal ( 175 wpm) and fast (225. 275 and 325 wpm) rates. in order to determine whether patients with left anterior temporal-lobe excisions arc un~~sually scnsitivc to the processing load of normal and accelcratcd speech. Because it seemed important to make the test stimuli as realistic as possible. only speech rates that could conceivably bc encountered in real life wcrc used, and speech was not compressed mechanically. If Icft anterior temporal-lobe rcscction does slow the rate at Lvhich speech can be proccsscd. one Lvould expect the comprehension and recall ofspoken material by patients with such Icsions to be facilitated by a slow rate of presentation. and increasingly hampcrcd as the rate of presentation increases. In contrast, comprehension by patients with right temporal-lobe excisions and by normal control subjects should not begin to decline until the presentation rate surpasses that of normal speech.
ME!THOD
Left temporal Right temporal Normal control
I e/-t~~np~~rc~/ lohe qwrp. The 11 patients111 this group (all right-handed) had each undergone a left anterior temporal-lobe rcsectwn. which always included the anterior temporal neocortex. the uncus and amygdala. and bchich In IOpat~enkencroached upon the hippocampus and parahippocampal gyrus. In four patients, hippocampal resection was limited to the pes of the hippocampus: in five patients the removal included up to 2.5 cm the body 01 the hippocampus: and in one patient. 3 cm of the hippocampus were excised. The primary auditory cortex (transverse gyri of Heschl) was spared in all LT patient\. Tahlc 2 describe!, the extent of removals along the SylLian lis\urc and along the base of the temporal lobe. Five of the patients In the LT group were seen 2 weeks after surgcr! and the remainder one or more years after their operation.
of
T;lblc
2 Wcchdcr
Group .____ I.& temporal Right remporal
Full-scale
IQ. Vcrhal
Full-scale IQ Mean Range 9X 106
xx 127 90 127
IQ.
Performance temporal-lobe
Verbal IQ Range MCUl 94 104
x0 129 x9 I22
IQ and extent or neocortical groups
Performance IQ Mean Range I05 I07
87 I.12 92 133
excision for subjects m
Extent of neocortial excision (cm) Babe of temporal Sylvlan lissul-e lobe Mean Range M&In Range 4.3 5.0
3 8 4.5 4.5 6.5
4.5 5.6
3.8 5.5 5.0 7.0
/(~y/~r-c~,~npc’~~~/ lohe qrr,up. Ten patients (six right-handed, three Icft-handed, nnd one of mixed handedwas) had a&ions from the right temporal lobe (RT), similar to those from the left hemisphere. Resection in one patient Inc,udcd part of primary audltory cortex. and nine of the patients had cxciaions of varying cstcnt l’rom the hlppocnmpus. .Thc hlppocampal Ircsection in two patients was llmited to the pes (I .5 cm): six remobala Included up 10 3.5 cm orlhc hippocampus; and one patient had had more than 2.5 cm ofthc hippocampus rcmo\ed. Extent ol’the removal\ along ~hc Sylvian fissure and along the base of the temporal lobe is described in Table 2. Twso 01‘these ipaticnts wcrc seen 2 weeks after surgery, one 4 months after surgery, and the remainder one or more yea)‘\ after their operation. Thcrc was a slgniticant diffcrcnce hethccn patient groups v,ith respect to the sire of the neocortical excision. The excisions were Iargcr in RT patients than in LT patients. both along the Sylcian fissure [F ( I. 20): IS 68.P
of
of
The coice of the cxpcrirncnler (I.J.) mas used in the digitization all the stimuli. The signals ucrc fed through 21 DI:RM high-hias liltel- and ;I TASCAM M-06 amplilier. and the stimuli were digitized at 16.000 samples per XC using MITSYN software running on :I Compaq 386 Model I20 with II Data Translation analogue suh5ystem. The :,tnnuli \\ere prcscntcd to the subjects through earphones (Sennheibcr) via the amplllier and the computer. using MAPLE soli\\are. All suhjccts performed t\\o tasks, in the order in which they are descl-ibed below. i.c\-,[,lr/-rl~,~,i,si(,,,/u\L. The 48 items making up this task were monoaqllahic utterances 3 5 phonemcs in length (24 words and 24 nonuords). selcctcd from thoac used by FRISK and MIL%\FK [7J. All the words had appro\lmatcly the \amc Crcqucnq of occIIrrcncc in the English lunguage [IO]. After each stimulus was heard. the aubjecth hit the key marked “T” with their dominant (preferred) hand ifthey thought that it has a real English {lord. and “F” wilh their Ilondominant hand if the! thought that it was ;i nonuord. Counterhalnncing ol’handswas not carried out because FKISS and MILKI K [7] had found in pilot testing that. when suhjecls ~crc required to USCtheir nondominant hand for a “true” response. manV more errors were made than when the dominant hand was used. Four practice trials L\crc prcbented lirst to fanliliarize the subjects with the task. The subject’s response and the latency to respond were recorded by the computer for each stimulus. The mtertrial tntcr\al (mcnsurcd from the instant ot” the suhject’b response to the onset of the next stimulus) was about 500 milliseconds (msec). .S~(~).~~-~~I~~~IO).\, /~,a!,. Tuenty-li\c stories. cuch follohcd by four statancnts (either true orfrdse) about the content of the stoky. wcrc used. Fil’twn stories uerc taken from FKISK and MILNN 177. The remaining IO stories were pretested on 20 normal control subjects in the following way: each story was recited to each subject at an average rate (about 225 wpm). and then the subject was required to decide Lbhcther each of the four statements that follwcd the star) ua$ true or falw. In accordance with the criterion set by FKISK and MII.KFK [7] to cnsurr that a statement was not
RESULTS The dependent
measurcs
were the number
or errors
made and latency
to respond.
Lcsic,rrl-tlrc,isit,il trdi Emus. A two-way analysis of variance with group as the between-subjects variable and stimulus type (word or nonword) as the within-subjects variable was pcrformed. There was no significant group-by-type interaction. but thcrc was a main effect of group [F (2. 32)=4.64. P~O.051. and of type of stimulus, [f-.(1, 33)=27.69, P
XI
RT
NC (“-13)
(n-lo,
Fig. I Lexicaldecision\. The pcrccntage ofwords and nonwords correctly identified by subjects in the left rcmporal-lobe (LT), right temporal-lobe (RT) and normal control (NC) groups, respectively.
175
225
275
325
Speech Rate (wpm) Fig. 2. Story recall: Mean percentage of true statements about stories correctly identiticd. as a function of speech rate. for subjects in the left temporal-lobe (LT). right temporal-lobe (RT) and normal control (NC) groups.
statcmcnts [F(4.12X)=14.17, P
1
175
225
275
326
Speech Rate (wpm) k’ig. 3. Story recall: Mean percentage of fake statements about stories corrcctlq dcntitkt. as ;L function of speech rate. for subjects in the left temporal-lobe (LT). right temporal-lobe (RT) and normal control (NC) groups.
wpm). The results of this ANOVA were similar to those from the ANOVA on all five r;ites. in that there were no significant interaction effects. but there was ;I main effect of group [F (2. 32)=6.35, 1-‘<0.00.5]. with normal control subjects making significantly fewer errors than patients in the LT group [Q (I.32)= 5.02, P
DISCUSSION The results of this experiment do not support the hypothesis th;tt left anterior temporal-lobe resection slows the rate at which speech can be processed. If the impairment in verbal memory demonstrated by patients with lesions of the left temporal lobe were in part attributable to slow processing of spoken language, then one would expect that, relative to the performance of normal subjects and patients with excisions from the right temporal lobe. comprehension of stories by such patients would have been disproportionately afrectcd by increasing the speech rate. Contrary to this hypothesis, no group-by-rate-of-presentation interaction was found in the number of errors made on the story-memory task. and no difference between the LT group and the other groups in the effect ofspeech rate on latency to respond to statements about story content was obtained. These results, taken together with those for visual presentation 171, make it extremely unlikely that abnormally slow processing
x3
r” 37
125
175
Speech Fig. 4. Story
oflanguage
22.5
275
325
Rate (wpm)
recall: Mean logarithmic response latency for correctly identified statements stories. for all subject groups combined. as a function or speech rate.
about
contributes significantly to the poor verbal memory of patients with anterior left temporal-lobe lesions. The LT group did commit more errors than the NC group when judging the correctness of both true and false statements about stories, at all speech rates. FRISK and MILNER [7] also observed that their LT subjects made more errors than NC subjects when judging the truth of statements about stories, and the converging results of the two studies, showing that patients with left temporal lesions are poor at retaining short prose passages that they havejust heard or read, further illustrate the impaired recall of verbal material already established for such patients [I I, 121. This deficit is probably the result of a retention impairment demonstrable when the information load exceeds the capacity of immediate memory. The finding that, for all groups. latencies to respond to both true and false statements shortened as the rate of presentation increased is different from the results for visual presentation. Thus, in the present experiment, response times were slower for material presented at an average rate (175 wpm) than for material presented at fast rates. In seeming contrast, FRISK and MILNER [7] found that subjects responded more quickly when the story preceding the statements was presented at the rate at which they were used to receiving it Inormal reading rate; 5 wps). There is no particular reason to expect a correspondence on this point. given that the processes involved in the perception ofspoken and written language are different. On the lexical-decision task, all subjects committed more errors when judging the legitimacy of nonwords than when judging the legitimacy of words. In addition. response latencies were longer for nonwords than for words. This latter finding is consistent with those of previous studies 17, 171, and seems to be a robust and general property of all “same” “difyerent” judgemcnts about codable stimuli [I, 41. The finding that the LT group made more errors than the NC group in judging the legitimacy of both words and nonwords was interesting. FRISK and MILNER [7] found no differences between their LT group and other groups in the performance of a lexical decision task in which the stimuli were presented visually. We are now examining this possible dissociation by performing a study in which normal control subjects and patients with unilateral anterior temporal-lobe resections perform lexical-decision tasks in both modalities, so that comparisons of performance between modalities can be made within subjects. Results of recent positron emission
tomographic studies suggest that different anatomical regions subserve visual and auditory single-word proccssing 18. 141, and our findings. if confirmed. would add further support to this hypothesis.
THE EFFECT OF PRESENTATION RATE ON THE COMPREHENSION AND RECALL OF SPEECH AFTER ANTERIOR TEMPORAL-LOBE RESECTION INGRIII JOHNSRUDE and BRENDA MILNER Department of Psychology and Department of Neurology and Neurosurgery, McGill Llniversity; and the Montreal Neurological Institute, 3801 University St, Montreal. Quebec, Canada H3A ?B4 (Rewired
I9 March
1993; rrccepted 7 July 1993)
Abstract-Twenty-two
paticnfs who had undergone unilateral anterior resection of either the left (LT) or right (RT) temporal lobe and 13 normal control (NC) subjects were tested on a lexicaldecision task and a story-memory task. Stimuli were presented aurally. and. in the latter task, at five different speech rates ranging from 125 words per minute (wpm) to 325 wpm. Memory for stories in the LT group was not abnormally sensitive to the effect of increasing rate. although it was inferior to that of the NC group at all speeds. This argues against the hypothesis that the verbal memory defcit evident after LT resection is partially attributable to an abnormally slow processing of verbal information. But LT subjects were impaired at judging the lexicality of uords and nonwords presented aurally, suggestmg that the left anterior temporal region plays a special role in the processing of speech sounds.
INTRODUCTION IT IS WELL ESTABLISHEDthat left anterior temporal-lobe resection impairs memory for verbal material. Indeed, impoverished memory for short prose passages (four or five sentences in length) remains, in our experience, the most conspicuous deficit exhibited by patients with such removals [6]. This deficit may, in part, be attributable to impaired processing of the material to be remembered: in particular, indirect evidence from several studies [ 12, 16, 191 indicates that patients with a left anterior temporal-lobe excision are impaired relative to those with right temporal-lobe excisions and normal subjects at assimilating and making use of information under time pressure. The question therefore arises as to whether, after left anterior temporal-lobe resection. patients are able to process verbal information as rapidly as normal subjects. This hypothesis was first suggested by BLAKEMOREand FALCONER [a], and was examined by FRISK and MILNER [7] in the visual modality. FRISK and MILNER [7] assessed comprehension of single words and short prose passages at three different presentation rates, using a rapid serial visual presentation (RSVP) procedure. For all groups, comprehension of stories decreased with increasing rate of presentation, and, although the left temporal-lobe group with large hippocampal removals made more errors than other groups when recognizing story content, their performance was not differentially impaired by increasing rate of presentation, relative to that of other groups. Despite these negative findings for visually presented material, the possibility remains that different results might be obtained in the auditory modality. The neural substrates of auditory and visual language processing are not identical [8, 143, and converging evidence
from several studies suggests that the left anterior temporal ncocortex plays ;I special role in the re-evocation of the sounds of recently heard words [ I3 161. Thus, excisions from this region may affect the rate at which spoken Innguagc is proccsscd. The prcscnt study-. dcsigncd to test this hypothesis, is an adaptation of FRISK and MILXR’S [7] procedure to the auditory modality. Normal English-speaking subjects find a speech rate of about 165 wpm the easiest to understand 191, and rutcs below 139 wpm and abovz 189 wpm arc judged to be “slow” and “fast”. respectively [S]. In the present study. sentences and stories wcrc prcscntcd aurally at slow ( I25 wpm) normal ( 175 wpm) and fast (225. 275 and 325 wpm) rates. in order to determine whether patients with left anterior temporal-lobe excisions arc un~~sually scnsitivc to the processing load of normal and accelcratcd speech. Because it seemed important to make the test stimuli as realistic as possible. only speech rates that could conceivably bc encountered in real life wcrc used, and speech was not compressed mechanically. If Icft anterior temporal-lobe rcscction does slow the rate at Lvhich speech can be proccsscd. one Lvould expect the comprehension and recall ofspoken material by patients with such Icsions to be facilitated by a slow rate of presentation. and increasingly hampcrcd as the rate of presentation increases. In contrast, comprehension by patients with right temporal-lobe excisions and by normal control subjects should not begin to decline until the presentation rate surpasses that of normal speech.
ME!THOD
Left temporal Right temporal Normal control
I e/-t~~np~~rc~/ lohe qwrp. The 11 patients111 this group (all right-handed) had each undergone a left anterior temporal-lobe rcsectwn. which always included the anterior temporal neocortex. the uncus and amygdala. and bchich In IOpat~enkencroached upon the hippocampus and parahippocampal gyrus. In four patients, hippocampal resection was limited to the pes of the hippocampus: in five patients the removal included up to 2.5 cm the body 01 the hippocampus: and in one patient. 3 cm of the hippocampus were excised. The primary auditory cortex (transverse gyri of Heschl) was spared in all LT patient\. Tahlc 2 describe!, the extent of removals along the SylLian lis\urc and along the base of the temporal lobe. Five of the patients In the LT group were seen 2 weeks after surgcr! and the remainder one or more years after their operation.
of
T;lblc
2 Wcchdcr
Group .____ I.& temporal Right remporal
Full-scale
IQ. Vcrhal
Full-scale IQ Mean Range 9X 106
xx 127 90 127
IQ.
Performance temporal-lobe
Verbal IQ Range MCUl 94 104
x0 129 x9 I22
IQ and extent or neocortical groups
Performance IQ Mean Range I05 I07
87 I.12 92 133
excision for subjects m
Extent of neocortial excision (cm) Babe of temporal Sylvlan lissul-e lobe Mean Range M&In Range 4.3 5.0
3 8 4.5 4.5 6.5
4.5 5.6
3.8 5.5 5.0 7.0
/(~y/~r-c~,~npc’~~~/ lohe qrr,up. Ten patients (six right-handed, three Icft-handed, nnd one of mixed handedwas) had a&ions from the right temporal lobe (RT), similar to those from the left hemisphere. Resection in one patient Inc,udcd part of primary audltory cortex. and nine of the patients had cxciaions of varying cstcnt l’rom the hlppocnmpus. .Thc hlppocampal Ircsection in two patients was llmited to the pes (I .5 cm): six remobala Included up 10 3.5 cm orlhc hippocampus; and one patient had had more than 2.5 cm ofthc hippocampus rcmo\ed. Extent ol’the removal\ along ~hc Sylvian fissure and along the base of the temporal lobe is described in Table 2. Twso 01‘these ipaticnts wcrc seen 2 weeks after surgery, one 4 months after surgery, and the remainder one or more yea)‘\ after their operation. Thcrc was a slgniticant diffcrcnce hethccn patient groups v,ith respect to the sire of the neocortical excision. The excisions were Iargcr in RT patients than in LT patients. both along the Sylcian fissure [F ( I. 20): IS 68.P
of
of
The coice of the cxpcrirncnler (I.J.) mas used in the digitization all the stimuli. The signals ucrc fed through 21 DI:RM high-hias liltel- and ;I TASCAM M-06 amplilier. and the stimuli were digitized at 16.000 samples per XC using MITSYN software running on :I Compaq 386 Model I20 with II Data Translation analogue suh5ystem. The :,tnnuli \\ere prcscntcd to the subjects through earphones (Sennheibcr) via the amplllier and the computer. using MAPLE soli\\are. All suhjccts performed t\\o tasks, in the order in which they are descl-ibed below. i.c\-,[,lr/-rl~,~,i,si(,,,/u\L. The 48 items making up this task were monoaqllahic utterances 3 5 phonemcs in length (24 words and 24 nonuords). selcctcd from thoac used by FRISK and MIL%\FK [7J. All the words had appro\lmatcly the \amc Crcqucnq of occIIrrcncc in the English lunguage [IO]. After each stimulus was heard. the aubjecth hit the key marked “T” with their dominant (preferred) hand ifthey thought that it has a real English {lord. and “F” wilh their Ilondominant hand if the! thought that it was ;i nonuord. Counterhalnncing ol’handswas not carried out because FKISS and MILKI K [7] had found in pilot testing that. when suhjecls ~crc required to USCtheir nondominant hand for a “true” response. manV more errors were made than when the dominant hand was used. Four practice trials L\crc prcbented lirst to fanliliarize the subjects with the task. The subject’s response and the latency to respond were recorded by the computer for each stimulus. The mtertrial tntcr\al (mcnsurcd from the instant ot” the suhject’b response to the onset of the next stimulus) was about 500 milliseconds (msec). .S~(~).~~-~~I~~~IO).\, /~,a!,. Tuenty-li\c stories. cuch follohcd by four statancnts (either true orfrdse) about the content of the stoky. wcrc used. Fil’twn stories uerc taken from FKISK and MILNN 177. The remaining IO stories were pretested on 20 normal control subjects in the following way: each story was recited to each subject at an average rate (about 225 wpm). and then the subject was required to decide Lbhcther each of the four statements that follwcd the star) ua$ true or falw. In accordance with the criterion set by FKISK and MII.KFK [7] to cnsurr that a statement was not
RESULTS The dependent
measurcs
were the number
or errors
made and latency
to respond.
Lcsic,rrl-tlrc,isit,il trdi Emus. A two-way analysis of variance with group as the between-subjects variable and stimulus type (word or nonword) as the within-subjects variable was pcrformed. There was no significant group-by-type interaction. but thcrc was a main effect of group [F (2. 32)=4.64. P~O.051. and of type of stimulus, [f-.(1, 33)=27.69, P
XI
RT
NC (“-13)
(n-lo,
Fig. I Lexicaldecision\. The pcrccntage ofwords and nonwords correctly identified by subjects in the left rcmporal-lobe (LT), right temporal-lobe (RT) and normal control (NC) groups, respectively.
175
225
275
325
Speech Rate (wpm) Fig. 2. Story recall: Mean percentage of true statements about stories correctly identiticd. as a function of speech rate. for subjects in the left temporal-lobe (LT). right temporal-lobe (RT) and normal control (NC) groups.
statcmcnts [F(4.12X)=14.17, P
1
175
225
275
326
Speech Rate (wpm) k’ig. 3. Story recall: Mean percentage of fake statements about stories corrcctlq dcntitkt. as ;L function of speech rate. for subjects in the left temporal-lobe (LT). right temporal-lobe (RT) and normal control (NC) groups.
wpm). The results of this ANOVA were similar to those from the ANOVA on all five r;ites. in that there were no significant interaction effects. but there was ;I main effect of group [F (2. 32)=6.35, 1-‘<0.00.5]. with normal control subjects making significantly fewer errors than patients in the LT group [Q (I.32)= 5.02, P
DISCUSSION The results of this experiment do not support the hypothesis th;tt left anterior temporal-lobe resection slows the rate at which speech can be processed. If the impairment in verbal memory demonstrated by patients with lesions of the left temporal lobe were in part attributable to slow processing of spoken language, then one would expect that, relative to the performance of normal subjects and patients with excisions from the right temporal lobe. comprehension of stories by such patients would have been disproportionately afrectcd by increasing the speech rate. Contrary to this hypothesis, no group-by-rate-of-presentation interaction was found in the number of errors made on the story-memory task. and no difference between the LT group and the other groups in the effect ofspeech rate on latency to respond to statements about story content was obtained. These results, taken together with those for visual presentation 171, make it extremely unlikely that abnormally slow processing
x3
r” 37
125
175
Speech Fig. 4. Story
oflanguage
22.5
275
325
Rate (wpm)
recall: Mean logarithmic response latency for correctly identified statements stories. for all subject groups combined. as a function or speech rate.
about
contributes significantly to the poor verbal memory of patients with anterior left temporal-lobe lesions. The LT group did commit more errors than the NC group when judging the correctness of both true and false statements about stories, at all speech rates. FRISK and MILNER [7] also observed that their LT subjects made more errors than NC subjects when judging the truth of statements about stories, and the converging results of the two studies, showing that patients with left temporal lesions are poor at retaining short prose passages that they havejust heard or read, further illustrate the impaired recall of verbal material already established for such patients [I I, 121. This deficit is probably the result of a retention impairment demonstrable when the information load exceeds the capacity of immediate memory. The finding that, for all groups. latencies to respond to both true and false statements shortened as the rate of presentation increased is different from the results for visual presentation. Thus, in the present experiment, response times were slower for material presented at an average rate (175 wpm) than for material presented at fast rates. In seeming contrast, FRISK and MILNER [7] found that subjects responded more quickly when the story preceding the statements was presented at the rate at which they were used to receiving it Inormal reading rate; 5 wps). There is no particular reason to expect a correspondence on this point. given that the processes involved in the perception ofspoken and written language are different. On the lexical-decision task, all subjects committed more errors when judging the legitimacy of nonwords than when judging the legitimacy of words. In addition. response latencies were longer for nonwords than for words. This latter finding is consistent with those of previous studies 17, 171, and seems to be a robust and general property of all “same” “difyerent” judgemcnts about codable stimuli [I, 41. The finding that the LT group made more errors than the NC group in judging the legitimacy of both words and nonwords was interesting. FRISK and MILNER [7] found no differences between their LT group and other groups in the performance of a lexical decision task in which the stimuli were presented visually. We are now examining this possible dissociation by performing a study in which normal control subjects and patients with unilateral anterior temporal-lobe resections perform lexical-decision tasks in both modalities, so that comparisons of performance between modalities can be made within subjects. Results of recent positron emission
tomographic studies suggest that different anatomical regions subserve visual and auditory single-word proccssing 18. 141, and our findings. if confirmed. would add further support to this hypothesis.