BRAIN INJURY, 1995, VOL. 9, NO. 6, 575-583

Depression among the head-injured and non-head-injured: a discriminant analysis

M . S . A L O I A j - $ , C . J . L O N G S a n d J . B . A L L E N $ tClinical Brain Disorders Branch, National Institute of Mental Health, St Elizabeths Hospital, Washington DC, USA $University of Mississippi, University, MS, USA §Memphis State University, Memphis, TN, USA

(Received 15 October 1994; accepted 12 Decembw 1994)

Neuropsychologists often use traditional psychological tests to assess depression following a head injury; but the assumption that depression with a head injury resembles that in an uniiijured person is suspect. The current study attempts to examine the cognitive manifestations of depression with and without a coexisting head-injury. Advanced statistical methods are used to assess whether or not the two depressions ‘look alike’ with respect to the neuropsychological sequelae of the disorders. A total of 1182 people were entered into one of two discriminant function analyses (DFA) for depression. Each person was a member of one of the following groups: (a) depressed, (bj non-depressed, (c) head-injured, or (d) head-injured and depressed. Two functions were performed for depression, one on the population ofhead-injured people and one on the population of uninjured people. Cross-validations were performed for each population and across populations in order to assess the utility of each population’s function for the opposite group. This comparison allows the researcher to indirectly compare depression in the two populations. Both functions were successfully applied to either population when MMPI variables were included in the analyses. However, when only cognitive variables were included the function performed on the non-head-injured population did not correctly classitjr head-injured people as depressed or non-depressed. One explanation for this is that the range ofcognitive scores in head-injured people is so great that it allows for a less accurate but more generalizable function. Suggestions for future research are discussed.

Introduction

Psychological depression is often manifested in lowered cognitive functioning. However, previous studies fail to generate consistent findings regarding the magni- tude of the depression-induced deficits. Further, there appears to be little agreement on the exact nature of the impairment, although some general cognitive declines are seen in functions such as memory [14], planning and executive functions [5-91, reaction time [8], lateralizing motor signs [lo], and timed cognitive tasks

Some believe depression affects particular types of memory as well as other specific functions assessed by neuropsychological tests [ 1,2 ,4 , 11,121. The research that supports these findings, however, is often based on poor methodology (e.g. sampling errors, misuse of statistical analyses, and pharmacological treatment effects as confounds) and is not easily replicated [4]. The lack of sound empirical information concerning the cognitive deficits ofdepressed individuals is an impediment to neuropsychologists seeking to assess loss of function in head-injured persons. The reason is that one of the

[7, 9, 101.

Correspondence to: Mark S. Aloia, MA, NIMH Neuroscience Center, St Elizabeths Hospital, 2700 Martin Luther King Jr Ave., SE Washington, DC 20032, USA.

0269-9052/95 $10.00 0 1995 Taylor k Francis Ltd

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

576 M. S . Aloia et al.

consequences of head injury is common depression. Thus, neuropsychologists who want to know the extent to which the deficits they observe are due directly to the physiological effects of the injury or indirectly to post-injury depression have no way of knowing. The two are confounded, and cannot be distinguished without further clarification of depression-induced deficits.

Two models of depression after head injury are shown in Figure 1. The first views the depression as remorse due to the loss ofcognitive competence. The second views both cogmtive deficits and depression as directly caused by the injury and in a reciprocal relationship. If we believe that depression is associated with a mourning over the loss of cognitive stabhty, then it would seem justifiable to expect the depression to subside as cognitive functioning improves. Research suggests that this does not occur [13]. It is believed that additional variables (e.g. awareness of deficits) are integral to the model.

The second model does not lend itselfwell to such generalizations. Since cognitive deficits associated with a head injury tend to improve with time but depression (as measured by the D scale of the Minnesota Multiphasic Personahty Inventory-MMPI) does not, and the two physiological processes feed each other in a reciprocal fashion (via awareness of deficit), we cannot yet determine the extent to which the remaining depression is due directly to the injury. It would, however, be advantageous for the neuropsychologist to be able to discriminate people who suffer fiom cognitive deficits associated with coexisting injury and depression, fiom those whose cognitive decline is associated with only one existing condition.

Aside firom the physiological comorbidrty, an additional problem arises in the assessment of the head-injured patient suffering fiom depression. Traditional tests are often employed by neuropsychologists in assessing personality and psychological concomitants of neurological impairments such as head injury [14]. The MMPI, for example, is often used in assessing head-injured people despite the fact that it was developed on a psychiatric population. Therefore, when interpreting the MMPI is it likely to expect head-injured people to have a greater probability of exhibiting psychopathological symptomatology, or should neuropsychologists develop a test more appropriate for measuring these traits in neurologically impaired populations? Indeed, the MMPI makes certain assumptions when applied to the neuropsychological population that may be unsuitable for this group [ 151. First, it assumes that this group wdl not respond differently due to fatigue when compared to the population on which it was normed. Second, it assumes that these people can read the questions accurately. Third, it assumes that they can comprehend the questions. Some of these administrative issues can be remedied methodologcally (e.g. by playing taped recordrngs ofthe questions). However, there is one problem that has not been well remedied. Some of the questions on the test assess neurological sequelae rather than measuring the clinical syndromes for which they were designed (see Table 1). For example, the depression @) scale contains questions appealing to a ‘slowed’ nature of the person, sleep problems, lack of concentration, or physical symptoms. These D-scale i tem can be sensitive to both neurological and to functional psychiatric disturbance. Questions ofthis sort do not load on the D scale alone [ 151. Consequently, it is hypothesized that if the head-injured person responds to such questions, s/he might present with a general elevated level ofpsychopathology. In a study conducted by Leininger, Kreutzer and Hdl[16] patients with a minor head injury were elevated on five of the 10 clinical scales of the MMPI. These elevations may be expected to correlate with the amount of neurologically somatic items loadrng on that particular scale. Therefore, many false positives for depression, as well as other psychatric disorders, might result when using this test to measure psychological hnctioning after head injury.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

Depression and head injury

1. Head Injury 0 Cognitive Deficits 0 Depression

577

Cognitive Deficits

a 2. Head Injury 8

c1

Depression

Figure 1 . Two proposed models ofthe comorbidity ofdepression and cognitive defints accompanying a head injury.

This may cause the neuropsychologist to ponder the nature of depression after head injury. In doing so, a logical outcome suggests that the concomitant tearing and shearing of neuronal fibres, which accounts for much of the diffuse damage seen with such an injury [17], might produce depressive symptomatology. This neuronal loss would most definitely be associated with a subsequent change in a variety ofneurotransmitter activities throughout the brain. If t h s is the case, depression after head injury might be different in nature, and certainly in aetiology, when compared to depression without neurological impairment.

It is hypothesized that the variables discriminating depression w i h n a head-injured population are of relatively similar importance within a non-head-injured population. In order to determine the degree of similarity between a head-injured population and an uninjured population in the manifestations of depression, a dscriminant hnction analysis was performed on each population for depression. The functions were assessed for their accuracy by cross-validation. They were also compared by attempting to apply each function to the opposite population for group separation (depressed vs. non-depressed). This cross-population cross-validation was tested twice: once with the inclusion of MMPI variables (other than the D scale) and once without. The comparison of these hnctions gives the neuropsychologist a better understandmg of the nature of depression after head injury as compared to depression experienced in a non-organic population. For example, the hnctions would be different ifthe scales ofthe MMPI that consist ofquestions tapping neurological symptomatology discriminate in the head-injured group but not in the other group, thereby suggesting that depression manifests itself differently in the two populations, and indirectly suggesting that depression is not the same in the two populations.

Methods

Subjects

Subjects were selected fiom a data pool consisting ofclients tested in a neuropsychological laboratory in Memphs, Tennessee. Selection was based on group membership. The subjects in the pool had been referred to the neuropsychological assessment laboratory by neurologists, caseworkers for the Department ofvocational Rehabilitation, attorneys, and as normal controls. Data were used fiom their full neuropsychological batteries. All subjects were given a summary of their testing along with recommendations for hrther assessmenthntervention. Full neuropsychological reports were given to the referring parties. In all, 1182 subjects were chosen.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

578

N=277

Age =35.2

D=84.6

M. S . Aloia et al.

N=471

Age= 3 1.5

D=55.9

Table 1 . D-scale responses that migbt also tap neurological sequelae

N=220

Age-38.5 r D=86.1

Questions Responses

True/False

F F F F F F F F F F F T T T T T

I have a good appetite I am about as able to work as I ever was I seldom worry about my health My judgement is better than it ever was I am in just as good physical health as most of my friends I seem to be about as capable and s m a r t as most others around me During the past few years I have been well most of the time

I am neither gaining nor losing weight I have never felt better in my life than I do now My memory seems to be alright I have d~fficulty in starting to do things I work under a great deal of tension I find it hard to keep my mind on a task or a job My sleep is fidul and disturbed I am afraid I am losing my mind

I have never had a fit or a convulsion

N=214

Age =38.3

D=56.4

The subject pool consisted of four clinically different groups (see Figure 2). Groups A and B were paired so as to allow for a discriminant hnction analysis (DFA) for depression (e.g. a depressed group and a non-depressed group). These two groups consisted of head-injured (HI) individuals as defined by the assessment outcome and clinical interview. The two groups were similar with regard to severity ofinjury as assessed by length of post-traumatic amnesia. Group A was also defined as depressed, having a T-score of 70 or higher on the depression (D) scale of the MMPI (Dahlstrom, personal communication), whereas group B was defined as not depressed. The DFA would allow for the prediction of group membership for depression given that the entire population is head-injured.

DEPRESSED

IES

io

-

YES N O

GROUP A GROUP B

GROUP C GROUP D

- No@. D=D-scale SWIC on the MMPI: D and Age an given as means.

Figure 2. Description of thefour clinical groups and their appropriate sample sixes.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

Depression and head injury 579

The second pair of groups, groups C and D, consisted of non-head-injured (NHI) indwiduals. Group C was also considered depressed as defined above, whereas group D was not depressed. This separation allowed for another DFA for group membership based on depression. In this case, however, the population was non-head-injured (NHI).

Materials

The neuropsychological test scores available in the data pool consisted of a modified Halstead-Reitan Neuropsychological Test Battery developed for use at this particular laboratory. Tests used in this battery, along with their reliability coefficients, are included in the Appendix [15, 181.

Design and procedure

All subjects were administered the tests in their standard form (e.g. according to the manual devised for the modified battery) by trained technicians [19]. Interpretations were made under the supervision of a practising clinical neuropsychologist certified by the American Board of Professional Neuropsychologists. The only variation to the original normative admmistration occurred with the grip strength and tapping subsections of the psychomotor testing. On these sections, subjects’ highest scores were used in place oftheir averages, in an attempt to obtain a measure of the subjects’ best possible performance.

The two groups, within each population, were analyzed using a hierarchical discriminant function analysis (stepwise method followed the entered hierarchy) of the SPSS-X mainfiame statistical package. Age, education and gender were controlled by entering them first in a hierarchical fashion so as to remove the variance accounted for by these variables. Hence, the remaining variables that emerge in the function do so in a stepwise manner fiom a pool of variance that reflects no differences in age, education or gender. The functions for depression were performed on 70% of each population and cross-vahdations were run on the remaining 30%. Adequacy was then determined by a statistical analysis of the hit rate of the original function with its cross-validation using chi-square analyses. Once adequacy was determined, a DFA for depression was performed on each entire population and cross-population cross-validations were run to assess the adequacy of each population’s equation to the opposite population. The equations were again tested to see if they were significantly different using chi-square analyses. Missing data were substituted with group means in concordance with recommended procedures [20].

Results

A discriminant function analysis for depression using the entire head-injured population was reliably successful (accuracy of its cross-vahdation did not differ fiom the original equation) at classifylng depressed and non-depressed individuals 85.83% of the time. An attempted cross-validation with the non-head-injured population was equally successful (85.64%), ~ ~ ( 1 ) = 0.00, p > 0.10. The DFA for depression run on the entire non-head injured population was reliably successful at classifjrlng the groups 92.40% of the time. Its cross-population cross-validation was equally successful (77.41%), ~ ~ ( 1 ) = 2.43, p > 0.10. This suggests that the manifestations of depression in the two populations are similar. The two DFAs with their hit rates and cross-population cross-validations are presented in Table 2.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

580 M . S . Aloia et al.

Table 2 . Variables disniminating depression in two entire clinical populations with cross-population cross-validations

Step enteredt Function

HI NHI

1 MMPI-PT 2 MMPI-HY 3 Cornell Mehcal index

5 MMPI-SI 6 Strength left hand

4 MMPI-PA

7 MMPI-SC 8 MMPI-PA 9 Trails C

10 Trails B 11 WAIS-R Information 12 WAIS-R Digit Span

14 TPT Memory 15 TPT L Blocks 16 WMS-Orientation 17 TPT Both Time 18 19 20 21 Hit rate: 85.33% Cross validation: 85.64%

13 MMPI-PD

- -

-

-

MMPI-PT Cornell Mehcal Index WMS-MQ MMPI-SI WAIS-R Digit Symbol WMS-Mental Concentration MMPI-HY MMPI-MA WAIS-R Similarities WAIS-R Picture Arranging Trails A Trails B WAIS-R Full-scale IQ Thurstone Word Fluency WMS-Digit Span WAIS-R Digit Span WMS-Logical Memory TPT Both Blccks TPT Total Blocks Strength right hand Aphasia Screening Test 92.30% 77.41%

tStep entered after hierarchy of age, educahon and gender

Given the similarity of MMPI scores across these populations (see Figure 3), further analysis of these variables is merited. However, these data were not adequate for item analysis; therefore, the above procedures were applied to the data with the exclusion of the MMPI variables to independently assess the cognitive manifestations of depression across populations.

The DFA for depression performed on the head-injured population was reliably successful at classification 67.5 1% of the time. Its cross-population cross-vahdation was equally successfd (54-61%), ~ ~ ( 1 ) = - 2.46, p > 0.10. The DFA for depression performed on the non-head-injured population was reliably successful at classifjlng the groups 8038% of the time, while its cross-population cross-validation was significantly less successful at classification (48.26%), ~ ' (1) = 1 3 . 1 6 , ~ > 0.001. This suggests that, with the removal of the MMPI variables, the purely cognitive manifestations of depression are less similar across these populations,

Discussion

The hypothesis that the relative importance of the variables discriminating the groups is similar in each population is supported when the MMPI variables are included. However, when examining neuropsychological variables alone the hypothesis is rejected. It is interesting to note that a function run on head-injured individuals, with or without MMPI variables, is equally accurate at classifjmg people on the depression variable

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

Depression and head injury 581

M I

C 0 r 20 e s o 40_L H8 D Hy Pd Mf Pa Pt Sc Ma Si

MMPI Clinical Scales

-H I ‘Normals *HID - D

HbHead-Injured, DmDepressed

Figure 3. MMPI profiles of the four clinical groups.

regardless of their neurological status (head-injured or non-head-injured). However, a function run on non-head-injured persons is viable in classifying both populations on the same variable only if it includes MMPI scores. Therefore, when cognitive and neuropsychological variables are examined alone, a DFA for depression is all-encompass- ing only if it is run on head-injured people. It is proposed that this applicabihty to another population is due to the diversity of deficits seen in head-injured people. This population manifests numerous cogmtive &&culties as it is more heterogeneous relative to the uninjured population. The heterogeneity of this group produces a greater range ofscores. This increased range allows for greater generahzability ofthe hnction performed on these scores, thus accounting for the better cross-population cross-validation. Although this hnction may be more generahzable, it may lack accuracy when compared to the hnction with a restricted range (as evidenced by the lower overall accuracy of the hnctions run on the head-injured population). A DFA is more specific if the population consists of uninjured people. This suggests that depression can look like a head injury (because a head injury can have so many different profiles) but, when using multivariate analyses such as these, it is more dlfficult for a head injury with an accompanying depression to mask itself solely as a depressive disorder. This non-specific profde ofthe head-injured person is most probably due to the difkseness of the injury and the wide range of severities involved in a compilation of such people.

These findings imply that neuropsychological variables associated with ‘depression’ (as defined here) are not the same in head-injuredpeople as they are in non-head-injured people. As the severity between the two head-injured groups is similar, the DFA stresses only those differences associated with the initial discriminating variable (the D scale). Therefore, when using the MMPI to assess depression in a head-injured population one must be aware that there are a dlfferent set ofvariables associated with ‘depression’ in this group than there are in an uninjured group. These findings support the notion that ‘depression’, as defined by a traditional test normed on uninjured people, has limited applicabihty to a head-injured population. Such application might lead to misdiagnosis and/or inappropriate treatment.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

582 M . S . Aloia et al.

One criticism of this research might be that the sheer number of variables used is likely to increase the likelihood of multicollinearity. Thus, some variables are likely not to dwriminate simply because they covary with other discriminating variables. This is a crucial point ifone intends to interpret the nature ofthe &scriminators. However, these authors intend only to apply an adequate set of discriminating variables, whether or not other variables might also discriminate, to the separation ofthese groups. It is likely that those variables that are multicollinear in one DFA are also multicollinear in the other DFA. Therefore, one would still expect the DFAs to be applicable across populations.

Additional research is needed to further delineate depression among the two populations. Future research would be best guided by the application of an item analysis of the MMPI among head-injured persons. Such an analysis might generate a subset of items with increased relevance for differentiating subtypes of depression in a neurological population. Further, application of this multivariate methodology using more specific group membership criteria (e.g. right-frontal injury as apseudodepression) might aid in the development ofclinical theory and in answering questions ofien facing the neuropsychol- oglst [21].

References 1. BRESLOW, R., KOCSIS, J. and BELKIN, B.: Memory deficits in depression: evidence utilizing

the Wechsler Memory Scale. Perceptual and Motor Skills, 51: 541-542, 1980. 2. DAGLEISH, T. and WATTS, F. N.: Biases of attention and memory in disorders of anxiety and

depression. Clinical Psychology Review, 10: 689-704, 1980. 3. MARTIN, M. and CLARK, D. M.: Selective memory, depression and response bias: an unbiased

response. Cognitive Therapy and Research, 10: 275-278, 1986. 4. JOHNSON, M. H. and MAGARO, P. A.: Effects of mood and severity on memory processes in

depression and mania. Psychologicaf Bulletin, 101: 28-40, 1987. 5. COLE, K. D. and SAIRT, S. H.: Psychological deficits in depressed medical patients.]ournal of

Nervous and Mental Diseme, 172: 150-155, 1984. 6. HARRIS, M., CROSS, H. and VANNIELJWICEFLK, R.: The effects of state depression, induced

depression and sex on the Finger Tapping and Tactual Performance tests. Clinical Neuropsychology, 3: 28-34, 1981.

7. LEES-HALEY, P. R. and FOX, D. D.: Neuropsychological a s e positives in litigation: Trail Making Test findings. Perceptual and Motor Skills, 70: 1379-1382, 1990.

8. NEWMAN, P. J. and SWEET, J. J.: The effects of clinical depression on the Luria Nebraska Neuropsychological Battery. InternationalJournal OfClinual Neuropsychology, 8: 109-1 14, 1 986.

9. SELIN, C. L. and GO'ITSCHALK, L. A.: Schizophrenia, conduct disorder, and depressive disorder: neuropsychological speech sample and EEG results. Perceptual and Motor Skills, 57: 427-444, 1983.

10. GOLDSTEIN, S. G., FILSKOV, S. B., WEAVER, L. A. et al.: Neuropsychological effects of electroconvulsive therapy. Journal of Clinical Psychology, 33: 798-806, 1977.

11. HERTEL, P. T. and RUDE, S. S.: Recalling in a state of natural or experimental depression. Cognitive Therapy and Research, 15: 103-127, 1991.

12. STERNBERG, D. E. and JARVIK, M. E.: Memory hnctions in depression. Archives of General Psychiatry, 33: 219-225, 1976.

13. LONG, C. J.A model of recovery to maximize the rehabilitation of individuals with head trauma. Journal ofHead Injury, 2: 19-28, 1991.

14. FORDYCE, D. J., ROUECHE, J. R. and PRIGATANO, G. P.: Enhanced emotional reactions in chronic head trauma patients.jouma1 of Neurology, Neurosurgery, and Psychiatry, 46, 620-624, 1983.

15. LEZAK, M. D.: Neuropsychological Assessment (Oxford University Press, New York), 1976. 16. LEININGER, B. E., KREUTZER, J. S. and HILL, M. R.: Comparison of minor and severe head

injury emotional sequelae using the MMPI. Brain Injury, 5: 199-205, 1991.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.

Depression and head injury 583

17. GOLDSTEIN, G., MCCUE, M., TURNER, S. M. et al.: An efficacy study ofmemory training with

18. SPREEN, 0. and STRAUSS, E.: A Compendium ofNeuropsychologica1 Tests (Oxford University

19. LONG, C. J.: Cortical Brain Assessment Lab: Summary oftest administration procedures, scoring and

20. TABACHNICK, B. G., and FIDELL, L. S.: Using Multivariate Statistics, 2nd edn (Harper & Row,

21. ADAMS, K. h4.: Linear discriminant analysis in clinical neuropsychological research Clinical

closed-head injury. Clinical Neuropsychologist, 2: 251-259, 1988.

Press, New York), 1991.

norms (Available &om Charles J. Long, Memphis State University), 1989.

New York), 1989.

Neuropsychologist, 1: 259-272, 1979.

Appendix

1. Finger Tapping-Sometimes called the Finger Oscillation Test, this test measures the number of taps made with the forefinger in a 10-second period over five consecutive trials (I = 0.58 to 0.93).

2. Grip Strength-This tests the strength a person exerts on a dynamometer (r = 0.52 to 0.96).

3. Wechsler Adult Intelligence Scale-Revised (excluding the information, arithmetic, picture completion, and object assembly subtests-(r > 0.88).

4. Tactual Performance Test-This test requires the subject to fit 10 blocks into a board, with appropriate cut-out shapes, blindfolded and within 10 minutes. The subject performs the task with hidher right, left, and both hands. S/He is then tested for memory of the shapes and their relative positions on the board (r = 0.55 to 0.91).

5. Trails A, B, C-The subject is given a sheet of paper containing 25 circles, each encompassing a number for Trials A, a letter for Trials C, and alternating numbers and letters for Trials B. The subject must properly connect the circles as quickly as possible (alternating 1-A-2-B.. .for Trials B; r = 0.66 to 0.94).

6. Wechsler Memory Scale-This is a test devised to assess the verbal, visual and associative memory of the subject as well as h d h e r mental status. Ths test includes both short- and long-term memory measures (r = 0.76 to 0.89).

7. Minnesota Multiphasic Personality Inventory-r = 0.51 to 0.92. 8. Cornell Medical Index-This self-report survey was designed to test the amount of

distress experienced by medical patients (r not available). 9. Thurstone Word Fluency-This test examines the subject’s ability to generate words

beginning with certain letters and following certain rules (e.g. no proper nouns) in a gwen amount of time (r not available).

10. Aphasia Screening Test-This test assesses the subject across a wide range of aphasic &ficulties (e.g. comprehension, repetition, fluency-r not available).

11. Seashore Rhythm Test-This test requires the subject to judge rhythm patterns as the same as, or different fi-om, previously presented patterns (r not available).

12. Speech Perception Test-This test assesses the subject’s ability to understand spoken nonsense words (r not available).

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Alb

erta

on

11/2

6/14

For

pers

onal

use

onl

y.