DEVELOPMENT OF HEMISPHERIC SPECIALIZATION FOR SPEECH PERCEPTION1 Gina Geffen2 (The Flinders University of South Australia)
INTRODUCTION
First language acquisition has been proposed to occur in a critical period from two years to puberty. Left hemisphere specialization for language is thought to first occur between ages three and five and increase thereafter, with the right hemisphere gradually performing less of the language functions until puberty when the adult degree of dominance by the left hemisphere is permanently established (Lenneberg, 1967). The clinical evidence for this was that lesions in the right hemisphere cause more language disturbance in children than in adults (Basser, 1962). However, this data has been reexamined (Krashen, 197 3 ), and in all cases of injury to the right hemisphere resulting in speech disturbance, the lesion was incurred before the age of five. In children injured after five years, the effects of right lesions are the same as in adults. This suggests that the development of language lateralization is complete by the age of five and that this coincides with completed acquisition of major aspects of language. Although the five years old has not attained full language competence (Chomsky, 1969), speech is internalized and used functionally to encode verbal items in recall tasks (Conrad, 1971 ). Dichotic listening tests using verbal items have been used to measure the degree of lateralization of language in normal adults and children. In adults, a small but consistent right ear advantage has been found and explained by left cerebral dominance for speech perception (Darwin, 1971; Studdert-Kennedy and Shankweiler, 1970). This explanation rests on evidence from animal and human studies. In cats, stimulation of both ears simultaneously produces occlusion of the ipsilateral sensory inputs by the contralateral pathways before reaching the cortex (Rosenzweig, 1951). Dichotic presentation of verbal items 1 This work was supported by grants from the Australian Research Grants Committee and the Flinders University Research Board. 2 The author is grateful to M. Sexton and I. Colley for testing and data analysis, and for the co-operation of the Blackwood Primary School and the Seacliff Kindergarten.
Cortex (1976) 12, 337-346.
338
G. Geffen
to a split brain patient resulted in total right ear recall with no left ear items being reported, whereas with monaural presentation, items from either ear were equally well reported (Sparks and Geschwind, 1968; Milner, Taylor and Sperry, 1968). Many developmental studies have used dichotic procedures to examine the age of onset and degree of language lateralization in different age groups. These studies fall into two categories, those using lists of dichotic digits or words for recall, and those which presented single pairs of consonant vowel (CV) syllables or single word pairs differing in one phoneme. The first group of studies (Geffner and Hochberg, 1971; Kimura, 1963; Knox and Kimura, 1970; Satz, Bakker, Teunissen, Goebbel and Van der Vlugt, 1975), can be criticized on the grounds that it is not clear whether the right ear advantage is due to perceptual asymmetry between the ears or report of right ear items first, with the consequence of greater forgetting of the left ear input. Moreover, individual results were not reported nor ceiling effects controlled. The right ear advantage obtained with dichotic digits varied from 10 to 16 percent and was consistent across age groups from four to nine years (Geffner and Hochberg, 1971; Kimura, 1963; Knox and Kimura, 1970). Reanalysis of this data with a scoring procedure that corrects for guessing and accuracy variation across age groups confirmed the lack of significant change in the right ear advantage from four to nine years (Krashen, 1973 ). On the other hand, an increase in ear asymmetry with age has been reported, with the right ear advantage reaching significance only at age nine (Satz et al., 1975). The recognition measure used in the second group of studies (Berlin, Hughes, Lowe-Bell and Berlin, 1972; Bryden and Allard, 1974; Knox and Kimura, 1970), where single pairs of stimuli were presented reduces memory load but does not eliminate it. Subjects wrote down the syllables they heard (Berlin et al., 1972), or pointed to one of a set of pictures after hearing a pair of words (Knox and Kimura, 1970), or indicated (actual response unspecified) which of two stimuli was perceived (Bryden and Allard, 197 4 ). Two items have to be held in memory and reported, which means that order of report still contaminates the findings. This procedure is particularly difficult for children five years old or younger, so experimenters have used vocal report with these age groups, leading to hazardous judgements of whether the child said "pa" or "da" (e.g., Berlin et al., 1972). Right ear advantages of 6 percent (Berlin et al., 1972), and 18 percent (Knox and Kimura, 1970), with no changes across the different age groups (5 to 15 years) were reported in two studies, whereas Bryden and Allard ( 197 4) found in children aged 6 to 14 a range of ear differences between 1 and 20 percent (mean 10 percent) with a significant right ear advantage in only the 12 and 14 year olds. The lack of agreement on the magnitude of the right ear advantage at different ages cannot be ascribed to differences in method since both methods produce conflicting results. Both recall and recognition procedures require cor-
Development of hemispheric specialization for speech
339
rection for order of report and where a vocal response has been used, this could itself produce a laterality difference. The present study used a monitoring procedure in which word lists were simultaneously presented. If a specified target word was detected the subject responded with the right hand to right ear targets and the left hand to left ear targets. A binaural as well as a dichotic condition was used to examine the contribution of hand differences. Words which sounded similar to the target words were included in the lists to define a noise distribution in order to apply a signal detection analysis to hit and false positive rates. This procedure minimizes the memory contribution as far as possible, corrects for guessing, and examines the contribution of response bias in ear asymmetry. Reaction times (RTs) were also measured.
MATERIALS AND METHOD
Subjects
The subjects were 40 right-handed children. Twenty-four school children were tested in two age groups, one 6.5- 7.5 years (mean 7.25 years}, and the other 10- 11 years (mean 10.8 years), with 12 subjects in each age group. Sixteen kindergarten children aged 4.8 - 5.3 years (mean 5.0 years) were also tested. The data of 12 of these subjects is reported, since four subjects in this group had d' = 0 for the left input in the dichotic condition, indicating random responding. These four subjects' data were omitted from further analyses. Males and females were equally represented in each age group. Handedness was determined by teachers' report and by requiring each subject to "draw something" (kindergarten sample) or "write your name" (school sample). Materials
The stimulus input in each condition consisted of a series of 120 word-pairs of one syllable common nouns synchronized for onset within 80 msec., equated for playback volume, and spoken at a rate of one pair per second. One word in each pair was spoken by a woman, and the other by a man. Each list of 120 word-pairs was made up of three kinds of words: (a) target words: there was one target word per list with a probability of occurrence of 0.167. (b) noise words: these words shared two phonemes in common with the target word and their probability of occurrence was 0.167. (c) irrelevant words: these had one or no phoneme in common with the target word and occurred with a probability of 0.667. Target, noise and irrelevant words were randomly distributed within each list with the following constraints: (a) no target occurred in the first or last four word-pairs of of a list; (b) the target occurred an equal number of times in each ear and in each voice; (c) the target word never occurred in successive pairs. (d) target words were paired with themselves (T + T), with noise words (T + N) and irrelevant words (T + I) on an equal number of occasions in each list.
340
G. Geffen
Apparatus
The lists were presented to the subject through stereo headphones (Akai-ASE95) played from a Revox A77 two-channel tape recorder. Channels 1 and 2 of a second Revox A77 tape recorder were used to re-record channees 1 and 2 of the stimulus input received by the subject. When either response button was depressed, an 8 kHz, 300 msec. tone was recorded (sound on sound) on the respective channels of the second tape recorder. A 20,350 Hz tone (beyond the human hearing range) had been superimposed at the onset of each target word on the stimulus tape. This was reproduced on the response recording which was subsequently fed through Schmitt relays and an electronic timer, one channel at a time, to obtain target detection rates and auditory-manual RTs. Procedure
The task involved monitoring a series of auditorily presented word pairs and pressing buttons every time the target word was heard. With the binaural presentation, both ears received identical input, and the subjects pressed a button with the left forefinger when they heard a target in one voice, and depressed the other button with the right forefinger when they heard the target in the other voice. With dichotic presentation, one ear received words spoken in the man's voice while the other ear received the words recorded in the woman's voice. Targets detected in the right ear were responded to with the subject's right hand and left ear targets with the left hand. Each subject was tested individually, seated next to the experimenter in front of the button panel and given the following general instructions: "You are going to hear some words in both ears. If you hear POT in this ear press this button. If you hear POT in the other ear press that button. If you hear POT in both ears, press both buttons." Similar instructions were given for the two voices. Before each condition, the subject was given several practice trials at responding to the target word presented on its own in each of the relevant voice/ ear combinations to familiarize him/her with the response required for each condition. Subsequently, a further set of practice trials was given consisting of 30 wordpairs from either the first or second half of each stimulus list. The practice trials were designed to control for word frequency and familiarity effects. Experimental trials followed the practice trials in blocks of 60 word-pairs, with two blocks for each condition. After each block of 60 word-pairs, the headphones were reversed to control for differences between the playback channels and/ or headphones. Dichotic and binaural conditions were performed by each subject in one session lasting 30 - 40 minutes. Experimental design
A between group comparison was used for the age variable (5, 7 and 10 year olds ), whereas within group comparisons were made on Condition (binaural and dichotic) and Input (left or right ear in the dichotic condition; man's or woman's voice in the binaural condition). The dependent variables were hit and false positive rates, yielding d' and ~ measures, and RT. The following counterbalancing procedures were carried out with respect to each age group: Two stimulus lists were used for the two conditions. Each list was used an equal number of times for each condition, and for every subject a
Development of hemispheric specialization for speech
341
different list was used for each condition. The order of conditions was controlled between subjects. Similarly, in the dichotic condition, half the subjects received the man's voice on the left ear and half on the right, and in the binaural condition, half the subjects responded to the man's voice with their left hand and half with their right. RESULTS
The five measures, hit rate, false positive rate, d', ~ and RT, were each subjected to a four-way analysis of variance.
Hit rate Hit rate increased with age, F (2, 33) = 38.4, p < .001. In the dichotic condition more right ear targets received responses. There were also more right than left hand responses in the binaural condition, F (1, 33) = 21.3, p < .00 1. The binaural and dichotic conditions did not differ in terms of hit rate, F < 1. None of the interactions with age were significant. The interaction of condition (dichotic vs. binaural) X side of response was significant, F (1, 33) = 6.93, p < .02. The left-right difference was greater in the dichotic condition (25% right side advantage) than in the binaural condition (10% right hand advantage). More subjects showed ear differences in the dichotic condition than hand differences in the binaural condition (Table I). TABLE I
Mean Percentage Targets Detected (Hit Rate) in Each Age Group and Condition
Dichotic
Binaural Age Left**
Right
Ss*
Left**
Right
Ss*
5
29 (20)***
45 (15)
7
25(8)
55 (22)
9
7
46 (15)
55 (11)
7
41 (20)
65(20)
9
11
69 (8)
77 (10)
8
63 (23)
84 (12)
10
* The number of subjects (N = 12 for each age group) showing a left-right difference in the same direction as the means. ** Left and right refers to the responding hand in the binaural condition and to the ear and hand in the dichotic condition. *** Standard deviations are given in parentheses.
Essentially the same results were obtained when pairs of target words which occurred simultaneously (T + T) were left out and responses to targets paired with noise and irrelevant words were analysed.
G. Geffen
342
False positive rate
This refers to the number of noise words responded to. The age effect was not significant, F (2, 33) = 1.04, p < .05. More noise words received responses in the binaural than the dichotic conditions, F ( 1, 33) = 9.25, p < .01. The left-right difference was not significant, F (1, 33) = 2.51, p < .05. None of the interactions were significant. D prime
d' increased with age, ,p (2, 33) = 22.65, p < .001. d' was higher in the dichotic than the binaural condition, F (1, 33) = 17.80, p < .001; and was higher for the right than the left side, F (1, 33) = 12.31, p < .01. Condition X Side was significant, F (1, 33) = 6.52, p < .05. As with hit rate, the difference between the left and right sides was greater in the dichotic than the binaural conditions, showing that the contribution of the hand difference is small compared to the ear difference (Table II). TABLE
II
Mean d' for Each Age Group Showing Left-Right Differences in Each Condition Binaural
Dichotic
Age Left** .71 (.57)***
5
Right
Ss*
1.14 (.62)
7
.70 (.48)
5
1.04 (.66)
1.85 (.70)
11
7
1.75 (1.03)
2.77 (.96)
10
7
1.12 (.49)
1.22 (.47)
11
1.59 (.39)
1.73 (.57)
*
**
Left**
Right
Ss*
1.41 (.67)
9
*** as in Table I.
Beta
None of the main effects, nor interactions, were significant. Reaction time
This measure showed a decrease with increasing age, F (2, 33) = 16.0 5, p < .001. The binaural vs. dichotic conditions did not differ, F (1,33) = 1.03, p < .05. Responses on the right were faster than those on the left, F (1, 33) = 13.83, p < .001. The interaction of Condition X Side of response failed to reach significance, F (1, 33) = 4.04, p < .053. It can be seen in Table III that the left-right difference in the binaural condition is smaller than in the dichotic condition for each age group. The overall mean difference
Development of hemispheric specialization for speech
343
(left-right) for the dichotic condition was 1.39 msec. compared to 67 msec. in the binaural condition. TABLE III
Mean RT (msec.) for Each Age Group and Condition Showing Left-Right Differences Binaural
Dichotic
Age Left**
5
Right
Left**
Right
Ss*
7
1318 (268)
1096 (342)
8
7
960(215)
889 (178)
8
984 (205)
873(191)
11
11
807 (124)
772 (130)
6
860 (113)
776 {107)
8
*
1185 (240)*** 1102 (206)
Ss*
**
*** as in Table I.
DISCUSSION
The use of a monitoring task in dichotic listening confirms the findings of dichotic recall and recognition studies (Geffner and Hochberg, 1971; Kimura, 196.3; Knox and Kimura, 1970; and Berlin et al., 1972), that a right ear advantage is established by the age of 4 years and does not increase with age. This results differs from the reports of an increasing right ear effect with age (Bryden and Allard, 1974; Satz et al., 1975). The nature of the right ear advantage has been clarified by the signal detection analysis which indicates that the right ear pathway is more sensitive to verbal stimuli. The d' measure showed a clear right ear advantage, whereas the criterion for responding (B) was the same for both ears in all three age groups. This perceptual asymmetry was confirmed by shorter RTs to right ear targets. The contribution of the right hand response to the ear difference can be measured by comparing the dichotic and binaural conditions. Since localization was held constant in the binaural condition, no ear advantage is possible and any difference in response to the two inputs must be produced by a hand difference. The magnitude. of this difference can be subtracted from the difference found in the dichotic condition, and the remainder reflects the ear difference for the processing of verbal input. The overall mean right side advantage was 25% (hit rate) in the dichotic condition, and the right hand advantage was 11% in the binaural condition. The right ear advantage is therefore 14%. Similar calculations can be made for the RT data, yielding an average of 72 msec. faster responses to right ear targets. If the right ear advantage reflects hemispheric specialization for language, then lateralization would appear to be fixed by at least 5 years of age.
344
G. Geffen
Krashen's ( 197 3) analysis of the clinical evidence indicated that the process of lateralization is coincident with language acquisition. Categorical perception of speech sounds has been found in six week old infants (Eimas, Siqueland, Jusczyk and Vigorito, 1971 ). However, the greater auditory evoked responses from the left hemisphere observed in response to verbal stimuli decreased with increasing age suggesting that specialization of the brain for language exists at birth (Molfese, 1972 ). With shorter, slower lists and a single response it should be possible to examine the ear difference in children as young as 2 years with a monitoring task. This would indicate whether lateralization increases with language acquisition, or precedes it, in normal children. An alternative explanation of the right ear advantage observed in 4 - 11 year olds is that it reflects a quantitative distribution of attention rather than hemispheric specialization for language (Geffen, Bradshaw and Nettleton, 1973; Treisman and Geffen, 1968). The attention explanation predicts a decrease in the magnitude of the ear advantage with age as more control over strategies of attending is gained. When adult subjects focus attention on one of two inputs, there is no ear advantage, but the unattended right ear is more sensitive to verbal targets than the unattended left ear (Treisman and Geffen, 1968 ). A 50 msec. advantage of the right ear, and a significantly greater number of targets (6.2%) detected by the right ear in a monitoring task with divided attention with adult subjects has been reported (Springer, 1971 ). This is a smaller effect than observed in the present study. The magnitude of the right ear advantage appears to vary with the processing demands of the task (Bartz, Satz and Fennell, 196 7; Satz, 1968; Satz, Achenbach, Pattishall and Fennell, 1965), which will change depending on the age of the subjects. This might explain the discrepant results of both Bryden and Allard ( 197 4) and Satz et al. (1975) who found an increasing laterality effect with age; their younger subjects may have been performing at too low a level (floor effect) to observe ear differences. The attention strategy would involve processing of right ear items first and more efficiently than left ear items and would be consistent with the RT, d' and hit rate data from the present study. However, this interpretation fails to explain why there is a right rather than a left ear advantage, nor does it account for the finding that certain aspects of the speech signal (e.g., initial consonants) are responsible for the right ear advantage rather than the entire speech signal (Studdert-Kennedy and Shankweiler, 1970). It also does not explain why non-verbal inputs (melodies, sonar signals, environmental noises) produce a left ear advantage (Chaney and Webster, 1965; Curry, 1967; Knox and Kimura, 1970). The available evidence suggests that the left hemisphere may be specialized for the extraction of linguistic features from speech signals (Studdert-Kennedy and Shankweiler, 1970) and that this is responsible for the observed right ear advantage. The predictive validity of the right ear advantage in establishing which side of the brain is specialized for language needs to be tested by
Development of hemispheric specialization for speech
345
giving the dichotic monitoring task to patients who are receiving unilateral intracarotid injection of amobarbital for this purpose.
SuMMARY
When lists of word pairs were presented simultaneously in a dichotic monitoring task to three groups of right-handed subjects aged 5, 7 and 11 years, more target words were responded to, and reaction times was shorter, when these occurred in the right rather than the left ear. The magnitude of the effect did not change with age. A signal detection analysis showed that the right ear advantage was due to greater sensitivity of the right ear, rather than a response bias. Hand effects were assessed separately by using binaural stimulation in which each ear received two inputs, one in a male voice and the other in a female voice. The right hand responded to target words in one voice, and the left hand to target words in the other voice. The contribution of a right hand superiority to the ear effect was found to be minimal. The effects of language acquisition and attention strategies on the right ear advantage are discussed. It was concluded that at 5 years of age the left hemisphere is specialized for the analysis of speech signals. This produces a right ear advantage in competitive verbal listening.
REFERENCES BARTZ, W.H., SATZ, P., and FENNELL, E. (1967) Grouping strategies in dichotic listening: The effects of instructions, rate and ear symmetry, "]. Exp. Psychol.," 74, 132-136. BASSER, L. (1962) Hemiplegia of early onset and the faculty of speech with special reference to the effect of hemispherectomy, "Brain," 85, 427-460. BERLIN, C., HuGHES, L., LowE-BELL, S., and BERLIN, H. (1972) Dichotic right ear advantages in children 5 to 13, "]. Acoust. Soc. Am.," 53, 368. BRYDEN, M., and ALLARD, F. (1974) Dichotic listening and the development of linguistic processes, in Hemispheric Asymmetry of Function, ed. by M. Kinsbourne, Tavistock, London. CHANEY, R.B., and WEBSTER, J.C. (1965) Information in certain multidimensional signals, U.S. Navy Electron. Lab. Rep. No. 1339, San-Diego, California. CHOMSKY, C. (1969) The Acquisition of Syntax in Children from 5 to 10, MIT Press, Cambridge, Mass. CoNRAD, R. (1971) The chronology of the development of covert speech in children, "Develop. Psychol.," 5, 398-405. CuRRY, F. (1967) A comparison of left-handed and right-handed subjects on verbal and nonverbal dichotic listening tasks, "Cortex," 3, 342-352. DARWIN, C.]. (1971) Ear differences in the recall of fricatives and vowels, "Quart. ]. Exp. Psychol.," 23, 46-62. EIMAS, P.D., SIQUELAND, E., JusCZYK, P., and VIGORITo, ]. (1971) Speech perception in infants, "Science," 171, 303-306. GEFFEN, G., BRADSHAW, ].L., and NETTLETON, N.C. (1973) Attention and hemispheric differences in reaction time during simultaneous audio-visual tasks, "Quart. ]. Exp. Psychol.," 25, 404-412. GEFFNER, D., and HocHBERG, I. (1971) Ear laterality performance of children from low and middle socio-economic levels on a verbal dichotic listening task, "Cortex," 74, 193-203. KIMURA, D. (1963) Speech lateralization in young children as determined by an auditory test, "]. Comp. Physiol. Psychol.," 56, 899-902. KNox, C., and KIMURA, D. (1970) Cerebral processing of nonverbal sounds in boys and girls, "Neuropsychologia," 8, 227-237. KRASHEN, S.D. (1973) Lateralization, language learning and the critical period: some new evidence, "Lang. Learn.," 23, 63-74. LENNEBERG, E. (1967) Biological Foundations of Language, Wiley, New York.
346
G. Geffen
MILNER, B., TAYLOR, L., and SPERRY, R. (1968) Lateralized suppression of dichotically presented digits after commissural section in man, "Science," 161, 184-186. MoLFESE, D.L. (1972) Cerebral asymmetry in infants, children and adults: auditory evoked responses to speech and music stimuli, "]. Acoust. Soc. Am.," 53, 363 (A). RosENZWEIG, M.R. (1951) Representation of the two ears at the auditory cortex, "Am. ]. Physiol.," 167, 147-158. SATZ, P. (1968) Laterality effects in dichotic listening, "Nature," 218, 277-278. - , AcHENBACH, K., PATTISHALL, E., and FENNELL, E. (1965) Order of report, ear asymmetry and handedness in dichotic listening, "Cortex," 1, 377-396. - , BAKKER, D.]., TEUNISSEN, ]., GoEBBEL, R., and VANDER VLUGT, H. (1975) Developmental parameters of the ear asymmetry: A multivariate approach, "Brain and Lang.," 2, 171-185. SPARKS, R., and GESCHWIND, N. (1968) Dichotic listening in man after section of neocortical commissures, "Cortex," 4, 3-16. SPRINGER, S. (1971) Ear asymmetry in a dichotic detection task, "Percept. Psychophysics," 10, 239-241. STUDDERT-KENNEDY, M., and SHANKWEILER, D. (1970) Hemispheric specialization for speech perception, "]. Acoust. Soc. Am.," 48, 579-594. TREISMAN, A.M., and GEFFEN, G. (1968) Selective attention and cerebral dominance in perceiving and responding to speech message, "Quart. ]. Exp. Psychol.," 20, 139-150.
Gina Geffen, School of Social Sciences, The Flinders University of South Australia, Bedford Park 5042, Australia.
INTRODUCTION
First language acquisition has been proposed to occur in a critical period from two years to puberty. Left hemisphere specialization for language is thought to first occur between ages three and five and increase thereafter, with the right hemisphere gradually performing less of the language functions until puberty when the adult degree of dominance by the left hemisphere is permanently established (Lenneberg, 1967). The clinical evidence for this was that lesions in the right hemisphere cause more language disturbance in children than in adults (Basser, 1962). However, this data has been reexamined (Krashen, 197 3 ), and in all cases of injury to the right hemisphere resulting in speech disturbance, the lesion was incurred before the age of five. In children injured after five years, the effects of right lesions are the same as in adults. This suggests that the development of language lateralization is complete by the age of five and that this coincides with completed acquisition of major aspects of language. Although the five years old has not attained full language competence (Chomsky, 1969), speech is internalized and used functionally to encode verbal items in recall tasks (Conrad, 1971 ). Dichotic listening tests using verbal items have been used to measure the degree of lateralization of language in normal adults and children. In adults, a small but consistent right ear advantage has been found and explained by left cerebral dominance for speech perception (Darwin, 1971; Studdert-Kennedy and Shankweiler, 1970). This explanation rests on evidence from animal and human studies. In cats, stimulation of both ears simultaneously produces occlusion of the ipsilateral sensory inputs by the contralateral pathways before reaching the cortex (Rosenzweig, 1951). Dichotic presentation of verbal items 1 This work was supported by grants from the Australian Research Grants Committee and the Flinders University Research Board. 2 The author is grateful to M. Sexton and I. Colley for testing and data analysis, and for the co-operation of the Blackwood Primary School and the Seacliff Kindergarten.
Cortex (1976) 12, 337-346.
338
G. Geffen
to a split brain patient resulted in total right ear recall with no left ear items being reported, whereas with monaural presentation, items from either ear were equally well reported (Sparks and Geschwind, 1968; Milner, Taylor and Sperry, 1968). Many developmental studies have used dichotic procedures to examine the age of onset and degree of language lateralization in different age groups. These studies fall into two categories, those using lists of dichotic digits or words for recall, and those which presented single pairs of consonant vowel (CV) syllables or single word pairs differing in one phoneme. The first group of studies (Geffner and Hochberg, 1971; Kimura, 1963; Knox and Kimura, 1970; Satz, Bakker, Teunissen, Goebbel and Van der Vlugt, 1975), can be criticized on the grounds that it is not clear whether the right ear advantage is due to perceptual asymmetry between the ears or report of right ear items first, with the consequence of greater forgetting of the left ear input. Moreover, individual results were not reported nor ceiling effects controlled. The right ear advantage obtained with dichotic digits varied from 10 to 16 percent and was consistent across age groups from four to nine years (Geffner and Hochberg, 1971; Kimura, 1963; Knox and Kimura, 1970). Reanalysis of this data with a scoring procedure that corrects for guessing and accuracy variation across age groups confirmed the lack of significant change in the right ear advantage from four to nine years (Krashen, 1973 ). On the other hand, an increase in ear asymmetry with age has been reported, with the right ear advantage reaching significance only at age nine (Satz et al., 1975). The recognition measure used in the second group of studies (Berlin, Hughes, Lowe-Bell and Berlin, 1972; Bryden and Allard, 1974; Knox and Kimura, 1970), where single pairs of stimuli were presented reduces memory load but does not eliminate it. Subjects wrote down the syllables they heard (Berlin et al., 1972), or pointed to one of a set of pictures after hearing a pair of words (Knox and Kimura, 1970), or indicated (actual response unspecified) which of two stimuli was perceived (Bryden and Allard, 197 4 ). Two items have to be held in memory and reported, which means that order of report still contaminates the findings. This procedure is particularly difficult for children five years old or younger, so experimenters have used vocal report with these age groups, leading to hazardous judgements of whether the child said "pa" or "da" (e.g., Berlin et al., 1972). Right ear advantages of 6 percent (Berlin et al., 1972), and 18 percent (Knox and Kimura, 1970), with no changes across the different age groups (5 to 15 years) were reported in two studies, whereas Bryden and Allard ( 197 4) found in children aged 6 to 14 a range of ear differences between 1 and 20 percent (mean 10 percent) with a significant right ear advantage in only the 12 and 14 year olds. The lack of agreement on the magnitude of the right ear advantage at different ages cannot be ascribed to differences in method since both methods produce conflicting results. Both recall and recognition procedures require cor-
Development of hemispheric specialization for speech
339
rection for order of report and where a vocal response has been used, this could itself produce a laterality difference. The present study used a monitoring procedure in which word lists were simultaneously presented. If a specified target word was detected the subject responded with the right hand to right ear targets and the left hand to left ear targets. A binaural as well as a dichotic condition was used to examine the contribution of hand differences. Words which sounded similar to the target words were included in the lists to define a noise distribution in order to apply a signal detection analysis to hit and false positive rates. This procedure minimizes the memory contribution as far as possible, corrects for guessing, and examines the contribution of response bias in ear asymmetry. Reaction times (RTs) were also measured.
MATERIALS AND METHOD
Subjects
The subjects were 40 right-handed children. Twenty-four school children were tested in two age groups, one 6.5- 7.5 years (mean 7.25 years}, and the other 10- 11 years (mean 10.8 years), with 12 subjects in each age group. Sixteen kindergarten children aged 4.8 - 5.3 years (mean 5.0 years) were also tested. The data of 12 of these subjects is reported, since four subjects in this group had d' = 0 for the left input in the dichotic condition, indicating random responding. These four subjects' data were omitted from further analyses. Males and females were equally represented in each age group. Handedness was determined by teachers' report and by requiring each subject to "draw something" (kindergarten sample) or "write your name" (school sample). Materials
The stimulus input in each condition consisted of a series of 120 word-pairs of one syllable common nouns synchronized for onset within 80 msec., equated for playback volume, and spoken at a rate of one pair per second. One word in each pair was spoken by a woman, and the other by a man. Each list of 120 word-pairs was made up of three kinds of words: (a) target words: there was one target word per list with a probability of occurrence of 0.167. (b) noise words: these words shared two phonemes in common with the target word and their probability of occurrence was 0.167. (c) irrelevant words: these had one or no phoneme in common with the target word and occurred with a probability of 0.667. Target, noise and irrelevant words were randomly distributed within each list with the following constraints: (a) no target occurred in the first or last four word-pairs of of a list; (b) the target occurred an equal number of times in each ear and in each voice; (c) the target word never occurred in successive pairs. (d) target words were paired with themselves (T + T), with noise words (T + N) and irrelevant words (T + I) on an equal number of occasions in each list.
340
G. Geffen
Apparatus
The lists were presented to the subject through stereo headphones (Akai-ASE95) played from a Revox A77 two-channel tape recorder. Channels 1 and 2 of a second Revox A77 tape recorder were used to re-record channees 1 and 2 of the stimulus input received by the subject. When either response button was depressed, an 8 kHz, 300 msec. tone was recorded (sound on sound) on the respective channels of the second tape recorder. A 20,350 Hz tone (beyond the human hearing range) had been superimposed at the onset of each target word on the stimulus tape. This was reproduced on the response recording which was subsequently fed through Schmitt relays and an electronic timer, one channel at a time, to obtain target detection rates and auditory-manual RTs. Procedure
The task involved monitoring a series of auditorily presented word pairs and pressing buttons every time the target word was heard. With the binaural presentation, both ears received identical input, and the subjects pressed a button with the left forefinger when they heard a target in one voice, and depressed the other button with the right forefinger when they heard the target in the other voice. With dichotic presentation, one ear received words spoken in the man's voice while the other ear received the words recorded in the woman's voice. Targets detected in the right ear were responded to with the subject's right hand and left ear targets with the left hand. Each subject was tested individually, seated next to the experimenter in front of the button panel and given the following general instructions: "You are going to hear some words in both ears. If you hear POT in this ear press this button. If you hear POT in the other ear press that button. If you hear POT in both ears, press both buttons." Similar instructions were given for the two voices. Before each condition, the subject was given several practice trials at responding to the target word presented on its own in each of the relevant voice/ ear combinations to familiarize him/her with the response required for each condition. Subsequently, a further set of practice trials was given consisting of 30 wordpairs from either the first or second half of each stimulus list. The practice trials were designed to control for word frequency and familiarity effects. Experimental trials followed the practice trials in blocks of 60 word-pairs, with two blocks for each condition. After each block of 60 word-pairs, the headphones were reversed to control for differences between the playback channels and/ or headphones. Dichotic and binaural conditions were performed by each subject in one session lasting 30 - 40 minutes. Experimental design
A between group comparison was used for the age variable (5, 7 and 10 year olds ), whereas within group comparisons were made on Condition (binaural and dichotic) and Input (left or right ear in the dichotic condition; man's or woman's voice in the binaural condition). The dependent variables were hit and false positive rates, yielding d' and ~ measures, and RT. The following counterbalancing procedures were carried out with respect to each age group: Two stimulus lists were used for the two conditions. Each list was used an equal number of times for each condition, and for every subject a
Development of hemispheric specialization for speech
341
different list was used for each condition. The order of conditions was controlled between subjects. Similarly, in the dichotic condition, half the subjects received the man's voice on the left ear and half on the right, and in the binaural condition, half the subjects responded to the man's voice with their left hand and half with their right. RESULTS
The five measures, hit rate, false positive rate, d', ~ and RT, were each subjected to a four-way analysis of variance.
Hit rate Hit rate increased with age, F (2, 33) = 38.4, p < .001. In the dichotic condition more right ear targets received responses. There were also more right than left hand responses in the binaural condition, F (1, 33) = 21.3, p < .00 1. The binaural and dichotic conditions did not differ in terms of hit rate, F < 1. None of the interactions with age were significant. The interaction of condition (dichotic vs. binaural) X side of response was significant, F (1, 33) = 6.93, p < .02. The left-right difference was greater in the dichotic condition (25% right side advantage) than in the binaural condition (10% right hand advantage). More subjects showed ear differences in the dichotic condition than hand differences in the binaural condition (Table I). TABLE I
Mean Percentage Targets Detected (Hit Rate) in Each Age Group and Condition
Dichotic
Binaural Age Left**
Right
Ss*
Left**
Right
Ss*
5
29 (20)***
45 (15)
7
25(8)
55 (22)
9
7
46 (15)
55 (11)
7
41 (20)
65(20)
9
11
69 (8)
77 (10)
8
63 (23)
84 (12)
10
* The number of subjects (N = 12 for each age group) showing a left-right difference in the same direction as the means. ** Left and right refers to the responding hand in the binaural condition and to the ear and hand in the dichotic condition. *** Standard deviations are given in parentheses.
Essentially the same results were obtained when pairs of target words which occurred simultaneously (T + T) were left out and responses to targets paired with noise and irrelevant words were analysed.
G. Geffen
342
False positive rate
This refers to the number of noise words responded to. The age effect was not significant, F (2, 33) = 1.04, p < .05. More noise words received responses in the binaural than the dichotic conditions, F ( 1, 33) = 9.25, p < .01. The left-right difference was not significant, F (1, 33) = 2.51, p < .05. None of the interactions were significant. D prime
d' increased with age, ,p (2, 33) = 22.65, p < .001. d' was higher in the dichotic than the binaural condition, F (1, 33) = 17.80, p < .001; and was higher for the right than the left side, F (1, 33) = 12.31, p < .01. Condition X Side was significant, F (1, 33) = 6.52, p < .05. As with hit rate, the difference between the left and right sides was greater in the dichotic than the binaural conditions, showing that the contribution of the hand difference is small compared to the ear difference (Table II). TABLE
II
Mean d' for Each Age Group Showing Left-Right Differences in Each Condition Binaural
Dichotic
Age Left** .71 (.57)***
5
Right
Ss*
1.14 (.62)
7
.70 (.48)
5
1.04 (.66)
1.85 (.70)
11
7
1.75 (1.03)
2.77 (.96)
10
7
1.12 (.49)
1.22 (.47)
11
1.59 (.39)
1.73 (.57)
*
**
Left**
Right
Ss*
1.41 (.67)
9
*** as in Table I.
Beta
None of the main effects, nor interactions, were significant. Reaction time
This measure showed a decrease with increasing age, F (2, 33) = 16.0 5, p < .001. The binaural vs. dichotic conditions did not differ, F (1,33) = 1.03, p < .05. Responses on the right were faster than those on the left, F (1, 33) = 13.83, p < .001. The interaction of Condition X Side of response failed to reach significance, F (1, 33) = 4.04, p < .053. It can be seen in Table III that the left-right difference in the binaural condition is smaller than in the dichotic condition for each age group. The overall mean difference
Development of hemispheric specialization for speech
343
(left-right) for the dichotic condition was 1.39 msec. compared to 67 msec. in the binaural condition. TABLE III
Mean RT (msec.) for Each Age Group and Condition Showing Left-Right Differences Binaural
Dichotic
Age Left**
5
Right
Left**
Right
Ss*
7
1318 (268)
1096 (342)
8
7
960(215)
889 (178)
8
984 (205)
873(191)
11
11
807 (124)
772 (130)
6
860 (113)
776 {107)
8
*
1185 (240)*** 1102 (206)
Ss*
**
*** as in Table I.
DISCUSSION
The use of a monitoring task in dichotic listening confirms the findings of dichotic recall and recognition studies (Geffner and Hochberg, 1971; Kimura, 196.3; Knox and Kimura, 1970; and Berlin et al., 1972), that a right ear advantage is established by the age of 4 years and does not increase with age. This results differs from the reports of an increasing right ear effect with age (Bryden and Allard, 1974; Satz et al., 1975). The nature of the right ear advantage has been clarified by the signal detection analysis which indicates that the right ear pathway is more sensitive to verbal stimuli. The d' measure showed a clear right ear advantage, whereas the criterion for responding (B) was the same for both ears in all three age groups. This perceptual asymmetry was confirmed by shorter RTs to right ear targets. The contribution of the right hand response to the ear difference can be measured by comparing the dichotic and binaural conditions. Since localization was held constant in the binaural condition, no ear advantage is possible and any difference in response to the two inputs must be produced by a hand difference. The magnitude. of this difference can be subtracted from the difference found in the dichotic condition, and the remainder reflects the ear difference for the processing of verbal input. The overall mean right side advantage was 25% (hit rate) in the dichotic condition, and the right hand advantage was 11% in the binaural condition. The right ear advantage is therefore 14%. Similar calculations can be made for the RT data, yielding an average of 72 msec. faster responses to right ear targets. If the right ear advantage reflects hemispheric specialization for language, then lateralization would appear to be fixed by at least 5 years of age.
344
G. Geffen
Krashen's ( 197 3) analysis of the clinical evidence indicated that the process of lateralization is coincident with language acquisition. Categorical perception of speech sounds has been found in six week old infants (Eimas, Siqueland, Jusczyk and Vigorito, 1971 ). However, the greater auditory evoked responses from the left hemisphere observed in response to verbal stimuli decreased with increasing age suggesting that specialization of the brain for language exists at birth (Molfese, 1972 ). With shorter, slower lists and a single response it should be possible to examine the ear difference in children as young as 2 years with a monitoring task. This would indicate whether lateralization increases with language acquisition, or precedes it, in normal children. An alternative explanation of the right ear advantage observed in 4 - 11 year olds is that it reflects a quantitative distribution of attention rather than hemispheric specialization for language (Geffen, Bradshaw and Nettleton, 1973; Treisman and Geffen, 1968). The attention explanation predicts a decrease in the magnitude of the ear advantage with age as more control over strategies of attending is gained. When adult subjects focus attention on one of two inputs, there is no ear advantage, but the unattended right ear is more sensitive to verbal targets than the unattended left ear (Treisman and Geffen, 1968 ). A 50 msec. advantage of the right ear, and a significantly greater number of targets (6.2%) detected by the right ear in a monitoring task with divided attention with adult subjects has been reported (Springer, 1971 ). This is a smaller effect than observed in the present study. The magnitude of the right ear advantage appears to vary with the processing demands of the task (Bartz, Satz and Fennell, 196 7; Satz, 1968; Satz, Achenbach, Pattishall and Fennell, 1965), which will change depending on the age of the subjects. This might explain the discrepant results of both Bryden and Allard ( 197 4) and Satz et al. (1975) who found an increasing laterality effect with age; their younger subjects may have been performing at too low a level (floor effect) to observe ear differences. The attention strategy would involve processing of right ear items first and more efficiently than left ear items and would be consistent with the RT, d' and hit rate data from the present study. However, this interpretation fails to explain why there is a right rather than a left ear advantage, nor does it account for the finding that certain aspects of the speech signal (e.g., initial consonants) are responsible for the right ear advantage rather than the entire speech signal (Studdert-Kennedy and Shankweiler, 1970). It also does not explain why non-verbal inputs (melodies, sonar signals, environmental noises) produce a left ear advantage (Chaney and Webster, 1965; Curry, 1967; Knox and Kimura, 1970). The available evidence suggests that the left hemisphere may be specialized for the extraction of linguistic features from speech signals (Studdert-Kennedy and Shankweiler, 1970) and that this is responsible for the observed right ear advantage. The predictive validity of the right ear advantage in establishing which side of the brain is specialized for language needs to be tested by
Development of hemispheric specialization for speech
345
giving the dichotic monitoring task to patients who are receiving unilateral intracarotid injection of amobarbital for this purpose.
SuMMARY
When lists of word pairs were presented simultaneously in a dichotic monitoring task to three groups of right-handed subjects aged 5, 7 and 11 years, more target words were responded to, and reaction times was shorter, when these occurred in the right rather than the left ear. The magnitude of the effect did not change with age. A signal detection analysis showed that the right ear advantage was due to greater sensitivity of the right ear, rather than a response bias. Hand effects were assessed separately by using binaural stimulation in which each ear received two inputs, one in a male voice and the other in a female voice. The right hand responded to target words in one voice, and the left hand to target words in the other voice. The contribution of a right hand superiority to the ear effect was found to be minimal. The effects of language acquisition and attention strategies on the right ear advantage are discussed. It was concluded that at 5 years of age the left hemisphere is specialized for the analysis of speech signals. This produces a right ear advantage in competitive verbal listening.
REFERENCES BARTZ, W.H., SATZ, P., and FENNELL, E. (1967) Grouping strategies in dichotic listening: The effects of instructions, rate and ear symmetry, "]. Exp. Psychol.," 74, 132-136. BASSER, L. (1962) Hemiplegia of early onset and the faculty of speech with special reference to the effect of hemispherectomy, "Brain," 85, 427-460. BERLIN, C., HuGHES, L., LowE-BELL, S., and BERLIN, H. (1972) Dichotic right ear advantages in children 5 to 13, "]. Acoust. Soc. Am.," 53, 368. BRYDEN, M., and ALLARD, F. (1974) Dichotic listening and the development of linguistic processes, in Hemispheric Asymmetry of Function, ed. by M. Kinsbourne, Tavistock, London. CHANEY, R.B., and WEBSTER, J.C. (1965) Information in certain multidimensional signals, U.S. Navy Electron. Lab. Rep. No. 1339, San-Diego, California. CHOMSKY, C. (1969) The Acquisition of Syntax in Children from 5 to 10, MIT Press, Cambridge, Mass. CoNRAD, R. (1971) The chronology of the development of covert speech in children, "Develop. Psychol.," 5, 398-405. CuRRY, F. (1967) A comparison of left-handed and right-handed subjects on verbal and nonverbal dichotic listening tasks, "Cortex," 3, 342-352. DARWIN, C.]. (1971) Ear differences in the recall of fricatives and vowels, "Quart. ]. Exp. Psychol.," 23, 46-62. EIMAS, P.D., SIQUELAND, E., JusCZYK, P., and VIGORITo, ]. (1971) Speech perception in infants, "Science," 171, 303-306. GEFFEN, G., BRADSHAW, ].L., and NETTLETON, N.C. (1973) Attention and hemispheric differences in reaction time during simultaneous audio-visual tasks, "Quart. ]. Exp. Psychol.," 25, 404-412. GEFFNER, D., and HocHBERG, I. (1971) Ear laterality performance of children from low and middle socio-economic levels on a verbal dichotic listening task, "Cortex," 74, 193-203. KIMURA, D. (1963) Speech lateralization in young children as determined by an auditory test, "]. Comp. Physiol. Psychol.," 56, 899-902. KNox, C., and KIMURA, D. (1970) Cerebral processing of nonverbal sounds in boys and girls, "Neuropsychologia," 8, 227-237. KRASHEN, S.D. (1973) Lateralization, language learning and the critical period: some new evidence, "Lang. Learn.," 23, 63-74. LENNEBERG, E. (1967) Biological Foundations of Language, Wiley, New York.
346
G. Geffen
MILNER, B., TAYLOR, L., and SPERRY, R. (1968) Lateralized suppression of dichotically presented digits after commissural section in man, "Science," 161, 184-186. MoLFESE, D.L. (1972) Cerebral asymmetry in infants, children and adults: auditory evoked responses to speech and music stimuli, "]. Acoust. Soc. Am.," 53, 363 (A). RosENZWEIG, M.R. (1951) Representation of the two ears at the auditory cortex, "Am. ]. Physiol.," 167, 147-158. SATZ, P. (1968) Laterality effects in dichotic listening, "Nature," 218, 277-278. - , AcHENBACH, K., PATTISHALL, E., and FENNELL, E. (1965) Order of report, ear asymmetry and handedness in dichotic listening, "Cortex," 1, 377-396. - , BAKKER, D.]., TEUNISSEN, ]., GoEBBEL, R., and VANDER VLUGT, H. (1975) Developmental parameters of the ear asymmetry: A multivariate approach, "Brain and Lang.," 2, 171-185. SPARKS, R., and GESCHWIND, N. (1968) Dichotic listening in man after section of neocortical commissures, "Cortex," 4, 3-16. SPRINGER, S. (1971) Ear asymmetry in a dichotic detection task, "Percept. Psychophysics," 10, 239-241. STUDDERT-KENNEDY, M., and SHANKWEILER, D. (1970) Hemispheric specialization for speech perception, "]. Acoust. Soc. Am.," 48, 579-594. TREISMAN, A.M., and GEFFEN, G. (1968) Selective attention and cerebral dominance in perceiving and responding to speech message, "Quart. ]. Exp. Psychol.," 20, 139-150.
Gina Geffen, School of Social Sciences, The Flinders University of South Australia, Bedford Park 5042, Australia.
