The Psychological Construct of Word Fluency

57, 394–405 (1997) BL971755

BRAIN AND LANGUAGE ARTICLE NO.

The Psychological Construct of Word Fluency R. M. RUFF University of California, San Francisco; University of California, San Diego

R. H. LIGHT AND S. B. PARKER University of California, San Diego AND

H. S. LEVIN University of Maryland, Baltimore Measures of word fluency have been convincingly linked in the literature to damage in the left prefrontal lobe region. Yet, a reduction in word fluency has also been reported with diffuse, multifocal and nonfrontal lobe damage. Despite the undisputed neuropsychological application of multiple word fluency measures, the psychological construct underlying this measure is not well understood. In a sample of 360 normal adults stratified by age, gender, and level of education, we found that auditory attention and word knowledge were among the most important determinants. With respect to memory, short-term memory was not significantly correlated, but long-term memory was. Despite these three determinants, a large share of the variance of the multiple regression was still not accounted for, which underscores the partial independence of word fluency per se. Thus, we propose a distinction between (1) poor word fluency secondary to deficient verbal attention, word knowledge, and/or verbal long-term memory and (2) impaired word fluency without these three areas concurrently affected. Based on a review of the literature, it seems likely that in the latter condition, the profile is more associated with prefrontal lobe impairment, versus in the former condition, diffuse multifocal or nonfrontal lobe factors can play a role.  1997 Academic Press

Word fluency measures are among the most sensitive for disclosing cerebral dysfunction. Reduced rates of word fluency have been observed in paThe first three authors were associated with UCSD during the time this project was conducted. The authors express their gratitude to Arthur Benton, who provided input for the integration of our data into a theoretical framework. We also appreciate his diligence in the editorial process. Address reprint requests to Ronald M. Ruff, Ph.D., Neurobehavioral Rehabilitation, St. Mary’s Hospital and Medical Center, 450 Stanyan Street, 4th Floor, San Francisco, CA 94117. 394 0093-934X/97 $25.00

Copyright  1997 by Academic Press All rights of reproduction in any form reserved.

PSYCHOLOGICAL CONSTRUCT OF WORD FLUENCY

395

tients with aphasic disorders and in patients with frontal lobe damage (Bruyer & Tuyumbu, 1980). To date probably the most convincing brain– behavioral link has been reported by Warkentin, Risberg, Nilsson, Karlson, and Graae (1991), who measured regional cerebral blood flow (rCBF) in 39 healthy volunteers during their performance of a word fluency test. The left dorsolateral prefrontal cortex had increased flow, thus confirming the evidence from the lesion studies that verbal fluency is associated with left prefrontal functioning. The neuropsychological application of word fluency tasks has been extended to a range of multifocal or diffuse neurological etiologies; examples include alcoholism (Hewett, Nixon, Glenn, & Parsons, 1991), traumatic brain injuries (Gruen, Frankle, & Schwartz, 1990), Alzheimer-type dementia (Hart, Smith, & Swash, 1988; Kertesz, Appell, & Fisman, 1986), epilepsy (Smith, Craft, & Collins, 1986; Hermann, Seidenberg, Haltiner, & Wyler, 1992), and multiple sclerosis (Swirsky-Sacchetti, Mitchell, Seward, Gonzales, Lublin, Knobler, & Field, 1992). Although the clinical neuropsychological application of word fluency is undisputed, its status as a psychological construct is not well understood. In most word fluency tasks, a specific letter of the alphabet is provided to the subject, who is asked to generate as many words as possible that begin with the identified letter of the alphabet. On a descriptive level, this appears to involve (1) immediate attention to initiate the generation of words, (2) an available word knowledge (i.e., semantic/lexical system) from which to select, (3) the ability to retrieve from verbal declarative memory, and (4) an executive ability to coordinate this process, including working memory to monitor performance and avoid breaking the rules (e.g., no proper nouns). One major aim of this study was to explore whether word fluency relates to any of these four determinants; each determinant will be discussed below. It is reasonable to assume that one determinant of word generation requires immediate verbal attention. Indeed, reduced word fluency has been reported after subcortical involvement such as a striato-capsular lesion in the dominant hemisphere (Kennedy & Murdock, 1990), which may imply among other possibilities that reduced immediate attention can play a role. To explore this relation, we evaluated how word fluency correlates with the Digit Span subtest of the WAIS-R and Seashore Rhythm Test, both of which can be considered to be measures of attention. Since word knowledge is necessary for fluency, it is reasonable to assume that a larger vocabulary would allow for a larger pool from which words could be selected. Thus, educational level may play a role. Evidence for the effects of education is found in the normative data for the Controlled Oral Word Association (COWA) test which indicate the need to adjust an individual’s score for educational level (Benton, Hamsher, & Sivan, 1994). The effect of word knowledge is also supported by the evidence that word fluency can vary with different letters. In the English language fewer

396

RUFF ET AL.

words begin with the letters J and U vs. F, S, P, and T (according to frequency counts from dictionaries or Thorndike–Lorge norms), and this in turn influences fluency rates (Borkowski, Benton, & Spreen, 1967). This supports the notion that word fluency is influenced by the vocabulary in existence and the level of knowledge thereof. Thus, the correlation between word fluency and the Vocabulary subtest of the WAIS-R was determined. Verbal memory may also subserve word finding or retrieval. Martin, Loring, Meador, and Lee (1990) found patients with primarily memory-related difficulties to be impaired on verbal fluency. They evaluated word fluency performances in 32 patients with unilateral temporal lobe epilepsy (TLE), and the left TLE patients gave a significantly inferior performance on both the pre- and postoperative assessments as compared to both the right TLE patients and the 25 control subjects (see also Hermann et al., 1992). Retrieval difficulties in patients with dementing disorders may also account for why patients with Alzheimer-type dementia or idiopathic Parkinson’s disease (Lees & Smith, 1983) demonstrate a reduction in word fluency without further clinical evidence of linguistic disabilities or aphasia. Finally, Fields (1990) has suggested that word fluency measures appear effective in identifying word finding problems in learning disabled children. Since word finding involves retrieval operations, we explored the relationship of word fluency to the immediate recall of the Wechsler Short Stories and to the longterm delay of the Selective Reminding Test. Executive and frontal lobe based dysfunctioning also are determinants which negatively influence word fluency (Perret, 1974). More specifically, a deficiency in shifting of cognitive set may play a major role. When generating words, three principle cognitive sets have been identified by Gruen et al (1990). These include same initial consonant–vowel syllable (see–seed), semantic association (ship–sail), or same initial consonant blend (tree–trick). Thus, with problems in executive functioning, the fluidity and flexibility within and between these three strategies may be compromised as part of a difficulty in shifting between cognitive sets. To compare fluid and flexible processing in an analogous test, we compared word fluency to the Figural Fluency Test (Ruff, Light, & Evans, 1997; Ruff, 1988), which evaluates fluency in the visuospatial mode. To examine the ability to abstract among word categories, COWA scores were correlated with the Similarities subtest of the WAIS-R. Finally, an error analysis was conducted to examine the frequency distributions between those who repeat the same words (perseverate) and those who produce no perseverations at all. The question of an age effect on error production was explored. The error rate on the COWA was also correlated with errors on other measures, such as with the perseverative errors committed on the figural fluency tests and with the intrusion errors made on the Selective Reminding Test.

PSYCHOLOGICAL CONSTRUCT OF WORD FLUENCY

397

TABLE 1 Sample of Participants Arranged by Age and Education Education 12 Years or less Age 16 to 25 to 40 to 55 to

24 39 54 70

13 to 15 Years

16 Years or More

Men

Women

Men

Women

Men

Women

15 15 15 15

15 15 15 15

15 15 15 15

15 15 15 15

15 15 15 15

15 15 15 15

METHOD

Subjects The total sample of 360 normal volunteers ranged in age between 16 and 70 years and in education from 7 to 22 years. All participants were native English-speaking individuals. About 65% of the participants resided in California, 30% resided in Michigan, and the rest resided on the eastern seaboard. Care was taken to ensure that the population was heterogeneous with respect to age and education; there were four age groups (see Table 1). To assess the test– retest reliability, five or more randomly selected subjects from each of the 12 cells (see Table 1) were retested after a 6-month delay (i.e., 30% of sample was retested). This interval was chosen because it typically represents the interval often selected for follow-up testing within a clinical setting for evaluating gains or deterioration of neuropsychological functioning. All participants were screened to exclude those with a positive history of psychiatric hospitalization, chronic polydrug abuse, or neurological disorders.

Procedure Word fluency. The word fluency measure selected for study was the Benton COWA Test. The examiner’s instructions for the test are as follows: ‘‘I am going to say a letter of the alphabet, and I want you to say as quickly as you can all the words you can think of which begin with that letter. You may say any word at all except proper names, such as names of people or places. So you would not say ‘‘Rochester’’ or ‘‘Robert.’’ Also, do not use the same words again with a different ending, such as ‘‘eat’’ and ‘‘eating.’’ ‘‘For example, if I say ‘‘S,’’ you could say, ‘‘sun,’’ ‘‘sit,’’ ‘‘shoe,’’ or ‘‘slow.’’ Can you think of other words beginning with the letter ‘‘S’’?’’ Wait for the subject to give a word, indicate if the word is correct, and ask the subject to give another word beginning with the letter ‘‘S’’. Once two appropriate words beginning with the demonstration letter are given, say, ‘‘That is fine. Now I’m going to give you another letter, and again say all the words beginning with that letter that you can think of. Remember, no names of people or places, just ordinary words. Also, if you should draw a blank, I want you to keep on trying until the time limit is up. You will have a minute for each one.’’ The first letter is C, and 1 min is allowed, and the same applies for the letters F and L of Version A. In Version B, the same procedure is used with the alternate letters P, R, and W. The record sheet provides numbered lines on which the subject’s responses can be entered. If the speed of production is too fast to permit verbatim recording, then a 1 sign should be entered. However, all incorrect responses should be entered verbatim. Many words have two or more meanings, and a repetition of the word is accepted only in those cases where the

398

RUFF ET AL.

subject definitely indicated an alternate meaning. If the patient produces one or more questionable responses (e.g., ‘‘frank,’’ which could represent a proper name), the association should simply be recorded, and the subject should not be interrupted. However, at the end of the 1 min period of association, the subject should be asked what was meant by this word. Slang terms are generally admissible, as well as certain foreign words (e.g., lasagna), as long as these words are listed as standard English.

Neuropsychological Tests Attention and concentration. Two test measures were selected to capture aspects of attention and concentration. First, we administered the Digit Span subtest of the WAIS-R (Wechsler, 1981). According to Lezak (1995), the WAIS-R Digit Span is ‘‘the format in most common use for measuring the span of immediate verbal recall’’ (p. 357). This test consists of random number sequences which the examiner reads out loud; in the first section, the individual is asked to recall multiple sequences of digits forward, and in the second part the digit sequences are to be recalled backward. In ascending order, the digit sequence is increased until the upper threshold forward and backward is determined. A combined score of the two parts was then converted into a scaled score which was used for our analyses. The second measure administered was the Seashore Rhythm Test (Seashore, Lewis, & Saetveit, 1960), which is one of the most widely used nonverbal sustained attention measures (Lezak, 1995). This test evaluates the individual’s ability to determine whether 30 sequences of rhythms presented in sequential order are the same or different. The number of correct identifications was analyzed in our study. Verbal memory. Two verbal memory measures were selected to capture immediate and long-term recall. First, we administered the Wechsler Short Stories from the Wechsler Memory Scale (Wechsler, 1945) to evaluate immediate memory. On this task, two short stories are read by the examiner, and the testee is asked to recall each story immediately afterward. Story A contains 24 memory units or ‘‘ideas’’ and Story B is composed of 22 units, which were combined for our analysis as a single measure of immediate verbal recall. The second measure administered was the Selective Reminding Test (Buschke & Fuld, 1974), which is composed of 12 simple words, which are to be learned across multiple trials (up to 12 trials). After each trial, structured feedback by the examiner is provided. Long-term memory was measured by the total number of words recalled after a 60-min delay (Ruff, Light, & Quayhagen, 1969). Word knowledge. The WAIS-R Vocabulary subtest (Wechsler, 1981), is among the most commonly used measures for evaluating vocabulary (Lezak, 1995). Thirty-six words must be defined by the testee. These answers were scored in a standardized fashion and the scaled scores were analyzed in our study. Executive functioning. With our two selected measures, we were only able to address limited aspects of executive functioning. To assess verbal concept formation, we selected the WAISR Similarities test (Wechsler, 1981). On this measure, the subject is asked to explain what each pair of words has in common. Standardized scoring procedures are provided, and the scaled score was analyzed. Nonverbal executive functioning was evaluated by the Ruff Figural Fluency Test (RFFT), which captures features analogous to the COWAT (Ruff, 1988; Ruff et al., 1984). The RFFT is composed of five sections, each using structured background for examining nonverbal response fluency. Each of the five parts contains 40 squares containing five dots. Parts 1–3 contain the identical dot positions; however, in Parts 2 and 3, two types of interference patterns are included. The dots on Parts 4 and 5 are asymmetrically positioned with all squares alike on each page. The patients are instructed to draw as many different patterns as possible by connecting two or more dots. Performances are scored for the number of unique designs and the number of repetitions (i.e., perseverations). All testing was carried out by trained technicians in single sessions, and in the identical order. We recognize that the above test selection is merely representative of the functions that are to be addressed.

PSYCHOLOGICAL CONSTRUCT OF WORD FLUENCY

399

RESULTS

The first set of analyses correlated the total COWA scores with selected neuropsychological test variables. In a second step, the importance among these test measures as they relate to the COWA was explored. Effects of Immediate Attention COWA scores were associated with both measures of auditory attention; the correlation with the total correct identifications of the Seashore Rhythm test was positive and significant (r 5 .33, p , .001), as was the correlation with the WAIS-R Digit Span (r 5 .45, p , .001). Effect of Verbal Memory COWA scores were significantly positively correlated with the combined immediate recall of the Wechsler Short Stories (r 5 .22, p , .001). COWA scores were also significantly associated with the 60-min delayed recall of 12 words previously learned on the Selective Reminding Test (r 5 .17, p , .001). Effect of Word Knowledge The COWA scores were significantly and positively correlated with the scale scores of the WAIS-R Vocabulary (r 5 .41, p , .001). Effect of Executive Functioning The RFFT, presumed to be a visual analog of the COWA, measures nonverbal fluency where the subject is asked to complete as many designs as possible within a configuration of dots. It was found that the COWA total score was significantly associated with the total number of designs produced on the RFFT (r 5 .24, p , .001). In addition, a positive correlation was also found for a verbal abstraction test, namely between the WAIS-R Similarities and the COWA (r 5 30, p , .001). Most Important Neuropsychological Contributors Since the correlations between COWA scores and the above selected neuropsychological test data were all significantly positive, the question was raised as to the unique contributions of these related tests in predicting COWA score variance. To determine the unique contributions computed as semipartial correlations, a multiple regression methodology was used where total COWA score was the outcome variable, and WAIS-R subtest scores of Vocabulary, Digit Span, and Similarities as well as the Seashore Rhythm raw score, the RFFT total designs, the delayed recall of the Selective Reminding test, and Wechsler Short Stories were the predictor variables. A

400

RUFF ET AL.

backward selection approach was used such that all predictors were entered into the regression equation, and the nonsignificant predictors were removed one by one until only significant predictors remained. Semipartial correlations were then computed from these results. In the initial equation, the least significant predictor was the Wechsler Short Stories Immediate Recall. This would indicate that the COWA scores did not share a significant amount of unique variance with short stories, and by inference, short-term verbal memory is a function independent of verbal fluency. Thus, the short stories were removed from the equation, and the regression was recomputed. At the next step it was found that the RFFT design was not a significant predictor of COWA scores, and it was removed from the equation. Finally, WAIS-R Similarities was also not a significant contributor, and it was removed from the equation. After inspecting the correlations, it was found that the RFFT designs shared variance with auditory attention, and it is likely that the significant zero-order correlation between RFFT scores and COWA scores was due to that moderating effect. A similar moderating effect was found with Similarities; however, there was a large association between Similarities and Vocabulary, and once this shared variance was accounted for in the regression equation, the Similarities variable offered no significant unique contribution to the equation. The final equation included Selective Reminding, WAIS-R Vocabulary, Seashore Rhythm Test raw score, and WAIS-R Digit Span as significant predictors of COWA scores. The multiple R2 between this set of predictors and the COWA was .25, F(4, 329) 5 28.13, p , .0001, indicating that 25% of the variance in COWA scores was predicted by this set of variables. In terms of unique variance accounted for, the largest semipartial correlation was that between COWA scores and WAIS-R Vocabulary (sr 5 .25, p , .001). The second largest was the semipartial correlation between COWA scores and WAIS-R Digit Span (sr 5 .20, p , .001). This was followed by the semipartial correlation between COWA scores and Seashore Rhythm score (sr 5 .14, p , . 005), and the semipartial correlation between COWA scores and delayed recall of the Selective Reminding Delayed Recall (sr 5 .10, p , .05). Based on these data, the most important contributor to scores on the COWA was word knowledge, as measured by WAIS-R Vocabulary. This was followed by immediate auditory attention, as measured by WAIS-R Digit Span, then sustained auditory attention, as measured by the Seashore Rhythm Test, and finally by delayed verbal memory, as measured by the Selective Reminding Test. Error Analysis In analyzing the frequency of repetitions or perseverations on the COWA, a majority (56%) of subjects produced no perseverations at all. The distribu-

PSYCHOLOGICAL CONSTRUCT OF WORD FLUENCY

401

TABLE 2 Rate of Perseveration on the Controlled Oral Word Association Test, Separately by Age Age (years)

Na

Percentage

16 25 40 55

29 42 47 42

32 47 52 47

a

to to to to

24 39 54 70

Number exhibiting at least one perservation.

tion of perseveration scores decreased logarithmically, with the highest number of perseverations being six, which occurred only once in the sample. Thus, the perseveration measure was dichotomized into those who did not perseverate vs. those who did. No significant differences were noted in the rate of perseveration due to gender or level of education. However, age moderated the perseveration rate; χ2 (3,N 5 360) 5 8.01, p , .05. No significant differences in perseveration rate were found between the age groups between 25 and 70; however, those ages 16–24 perseverated at a significantly lower rate (32%) than did the combined age groups from 25 to 70 (49%); χ2 (1,N 5 360) 5 7.26, p , .01. Table 2 presents the perseveration rates across age levels. Are the perseverative errors on the COWA related to other measures of perseveration? There was no significant association between COWA perseveration and perseverations on the RFFT. However, COWA perseverations were significantly positively associated with the intrusion errors on the Selective Reminding Test (r 5 .20, p , .001); intrusion errors are words provided by the subject which are not part of the original 12-word list. DISCUSSION

The major aim of this paper was to explore the construct validity of COWA, a measure of word fluency. In a sample of normal adults COWA was most strongly correlated, according to semipartial correlations, with a measure of vocabulary or word knowledge and two measures of auditory attention. Moreover, COWA was also correlated with a measure of verbal memory, however, not short-term memory as measured by the Wechsler Short Stories, but with the long-term measure of the Selective Reminding Test. Therefore, our data suggest that word fluency is subserved by an interrelationship between attention, long-term verbal memory, and word knowledge. Although our hypothesized determinants are confirmed by significant correlations, most of the zero order correlations were less than .35. Even the multiple regression accounted for only 25% of the variance. Presented in

402

RUFF ET AL.

this context, although it is acknowledged that there may be other abilities that correlate with the COWA, it is probable that we have selected the most likely determinants. Our study further allowed us to identify certain variables that do not correlate significantly with the COWA. That is, perseverative errors on the COWA did not correlate with perseverative errors on the RFFT. As mentioned earlier, short-term memory also is not a determinant for the fluency rate of the COWA. However, since (1) we appear to have identified the most likely determinants and (2) a large share of the variance of the multiple regression was still not accounted for, the partial independence of the COWA as a capacity per se is underscored. This capacity does not imply functional significance, but rather refers to the unique aspect that is captured by the tests. Thus, the COWA measures a cognitive capacity to generate words from their initial letters; on the other hand, performance of this task emanates from a convergence of other cognitive abilities. This combination of a multifactorial construct and a partial independence of the COWA allows for a parsimonious explanation with respect to the neuropsychological findings. The notion of multiple determinants explains why patients with a range of focal lesions as well as more diffuse brain damage can demonstrate impaired performances of word fluency. Thus, if verbal attention, word knowledge, and/or verbal long-term memory are deficient, then word fluency can be affected as a secondary consequence. Patients with traumatic brain injury, Alzheimer-type dementia, epilepsy, or multiple sclerosis who suffer from a range of multifocal or diffuse damage are known to be compromised in the area of attention and/or memory, and this in turn limits word fluency. However, COWA also reflects a partially independent function supported by the fact that word fluency per se is linked to the left dorsolateral prefrontal cortex. This raises the question of the interpretive significance of COWA within a neuropsychological test battery. We acknowledge that the cognitive constructs contributing to COWA performances may be quite different between normal and brain-damaged individuals. Pendleton, Heaton, Lehman, & Hulihan (1982) analyzed a comparable word fluency measure, namely the Thurstone Word Fluency Test (TWFT; Thurstone, 1938; Thurstone & Thurstone, 1949) in a study including both brain damaged groups and a normal control group. The TWFT was sensitive to cerebral dysfunctioning regardless of whether the lesion was diffuse or focal. Consistent with earlier findings (e.g., Perret, 1974), Pendleton et al. found that fluency rates of the TWFT were more significantly impaired when the focal lesions involved the frontal lobes than when they involved nonfrontal lesions. However, when using step-wise discriminant function analysis with the TWFT scores alone, approximately 60–65% of the patients with focal cerebral lesions were correctly classified with regard to the presence or absence of frontal involvement. This ‘‘hit rate’’ was interpreted as insufficiently high to justify the TWFT alone as a ‘‘sign’’ of frontal impairment. Pendleton et al. further concluded that al-

PSYCHOLOGICAL CONSTRUCT OF WORD FLUENCY

403

though word fluency measures are a valuable addition to standardized neuropsychological test batteries, word fluency performances are best interpreted relative to other test performances. Based on our findings, this comparison among neuropsychological test performances should in the future include a distinction between (1) poor word fluency combined with deficient verbal attention, word knowledge, and/or verbal long term memory and (2) impaired word fluency without these three areas concurrently affected. Given this distinction, future studies may demonstrate a sufficient hit rate to justify the latter test profile as a sign of (predominantly left) prefrontal impairment. Prefrontal pathology has frequently been described as demonstrating a dissociation between knowing and doing (Luria, 1975; Milner, 1964; Teuber, 1964). This thought–action dissociation has been further described by Pribram (1971) as a disconnection of feedback–feedforward systems which in turn does not allow the knowledge to be adequately utilized. As Stuss and Benson (1984) pointed out, it is important to recognize that frontal lobe abnormalities affect cognitive functioning differently than lesions in other cortical areas. Thus, the execution of prefrontal lobe processing can be compromised (1) indirectly by nonfrontal deficits in auditory attention, long-term memory, and/or word knowledge or (2) directly by frontal lobe damage which results in an interference between action and knowledge. This action– knowledge difficulty can impair the ability to produce words based on (1) establishing sets or strategies of searching for words, (2) maintaining such strategies, and/or (3) flexibly and fluidly alternating between strategies. This may also account for the finding that patients with right, as well as left, frontal lobe disease show impairment on this verbal task (Benton, 1968; Ramier & He´caen, 1970). REFERENCES Benton, A. L. 1968. Differential behavioral effects in frontal lobe disease. Neuropsychologia, 6, 53–60. Benton, A. L., Hamsher, K. de S., & Sivan, A. B. 1994. Multilingual Aphasia Examination. Iowa City, IA: AJA Associates. Borkowski, J. G., Benton, A. L., & Spreen, O. 1967. Word fluency and brain damage. Neuropsychologia, 5(2), 135–140. Buschke, H., & Fuld, P. A. 1974. Evaluating storage, retention, and retrieval in disorders of memory and learning. Neurology, 24, 1019–1025. Bruyer, R., & Tuyumbu, B. 1980. Fluence verbale et le´ sions du cortex cerebral: Performances et types d’erreurs. L’Ence´ phale, 6(3), 287–297. Fields, M. R. 1990. A study of selected fluency measures to identify word finding deficits in learning-disabled children (Southern Connecticut State University, 1990). Masters Abstracts, 28(03), 424. Gruen, A. K., Frankle, B. C., & Schwartz, R. 1990. Word fluency generation skills of headinjured patients in an acute trauma center. Journal of Communication Disorders, 23(3), 163–170. Hart, S., Smith, C. M., & Swash, M. 1988. Word fluency in patients with early demential of Alzheimer type. The British Journal of Clinical Psychology, 27(2), 115–124.

404

RUFF ET AL.

Hermann, B. P., Seidenberg, M., Haltiner, A., & Wyler, A. R. 1992. Adequacy of language function and verbal memory performance in unilateral temporal lobe epilepsy. Cortex, 28, 423–433. Hewett, L. J., Nixon, S. J., Glenn, S. W., & Parsons, O. A. 1991. Verbal fluency deficits in female alcoholics. Journal of Clinical Psychology, 47(5), 716–720. Kennedy, M., & Murdock, B. E. 1990. Persistent word fluency deficit in patients with dominant hemisphere striato–capsular lesions: Further evidence for a subcortical role in language? Australian Journal of Human Communication Disorders, 18(2), 3–20. Kertesz, A., Appell, J., & Fisman, M. 1986. The dissolution of language in Alzheimer’s disease. Canadian Journal of Neurological Science, 13, 415–418. Lees, A. J. & Smith, E. 1983. Cognitive deficits in the early stages of Parkinson’s disease. Brain, 106(2), 257–270. Lezak, M. D. 1995. Neuropsychological Assessment. New York: Oxford Univ. Press. 3rd ed. Luria, A. R. 1973. The working brain (B. Haigh, Transl.). New York: Basic Books. Martin, R. C., Loring, D. W., Meador, K. J., & Lee, G. P. 1990. The effects of lateralized temporal lobe dysfunction on formal and semantic word fluency. Neuropsychologia, 28(8), 283–289. Milner, B. 1964. Some effects of frontal lobectomy in man. In J. M. Warrent & K. Akert (Eds.), The frontal granular cortex and behavior. New York: McGraw-Hill. Pp. 313– 334. Perret, E. 1974. The left frontal lobe of man and the suppression of habitual responses in verbal categorical behavior. Neuropsychologia, 12, 323–330. Pendleton, M. G., Heaton, R. K., Lehman, R. A., & Hulihan, D. 1982. Diagnostic utility of the Thurstone Word Fluency Test in neuropsychological evaluations. Journal of Clinical Neuropsychology, 4(4), 307–317. Pribram, K. H. 1971. Languages of the brain: Experimental paradoxes and principles in neuropsychology. Englewood Cliffs, NJ: Prentice-Hall. Ramier, A. M., & He´caen, A. 1970. Roˆle respectif des atteintes frontales et de la late´ralisation le´sionnelle dans les de´ ficits de la ‘‘fluence verbale.’’ Revue Neurologique, 123, 17– 22. Ruff, R. M. 1988. Ruff figural fluency test administrative manual. Odessa, FL: Psychological Assessment Resources, Inc. Ruff, R. M., Light, R. H., & Quayhagen, M. 1989. Selective Reminding Tests: A normative study of verbal learning in adults. Journal of Clinical and Experimental Psychology, 11, 539–550. Ruff, R. M., Light, R. H., & Evans, R. W. 1987. The Ruff Figural Fluency Test: A normative study with adults. Developmental Neuropsychology, 3, 37–51. Seashore, C. E., Lewis, D., & Saetveit, D. L. 1960. Seashore measures of musical talents— Revised. New York: Psychological Corporation. Smith, D. B., Craft, B. R., & Collins, J. 1986. Behavioral characteristics of epilepsy patients compared with normal controls. Epilepsia, 27(6), 760–768. Stuss, D. T., & Benson, D. F. 1984. Neuropsychological studies of the frontal lobes. Psychological Bulletin, 95(1), 3–28. Swirsky-Sacchetti, T., Mitchell, D. R., Seward, J., Gonzales, C., Lublin, F., Knobler, R., & Field, H. L. 1992. Neuropsychological and structural brain lesions in multiple sclerosis: A regional analysis. Neurology, 42(7), 1291–1295. Teuber, H. L. 1964. The riddle of frontal lobe function in man. In J. M. Warren & Y K. Alkert (Eds.), The frontal granular cortex and behavior. New York: McGraw-Hill. Pp. 410– 477. Thurstone, L. L. 1938. Primary mental abilities. Chicago: Univ. of Chicago Press. Thurstone, L. L. & Thurstone, T. 1949. Examiner manual for the SRA Primary Mental Abilities Test. Chicago: Science Research Associates. Warkentin, S., Risberg, J. Nilsson, A., Karlson, S. & Graae, E. 1991. Cortical activity during

PSYCHOLOGICAL CONSTRUCT OF WORD FLUENCY

405

speech production: A study of regional cerebral blood flow in normal subjects performing a word fluency task. Neuropsychiatry, Neuropsychology, Behavioral Neurology, 4(4), 305–316. Wechsler, D. A. 1945. A standard memory scale for clinical use. Journal of Psychology, 19, 87–95. Wechsler, D. A. 1981. Manual for the Wechsler Adult Intelligence Scale—Revised. New York: Psychological Corporation.