Professional Documents
Culture Documents
Troyer (2000)
Troyer (2000)
Troyer (2000)
To cite this article: Angela K. Troyer (2000): Normative Data for Clustering and Switching on Verbal Fluency Tasks, Journal
of Clinical and Experimental Neuropsychology, 22:3, 370-378
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss, actions,
claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Journal of Clinical and Experimental Neuropsychology 1380-3395/00/2203-370$15.00
2000, Vol. 22, No. 3, pp. 370-378 © Swets & Zeitlinger
ABSTRACT
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
Normative data for clustering and switching on verbal fluency tasks are provided. Four hundred and eleven
healthy adults between the ages of 18 and 91 were given tests of phonemic fluency (FAS or CFL) and
semantic fluency (Animals and Supermarket). Raw scores were corrected for demographic (i.e., age, edu-
cation, and sex) and test (i.e., fluency form) variables that were determined to make sizable contributions
to fluency performance. These normative data should be useful for clinicians and researchers in determin-
ing the nature of the fluency impairment in any given individual.
Tests of verbal fluency or word-list generation fluency, and is thought to be a relatively auto-
are frequently used in clinical and experimental matic process. Switching involves cognitive
examinations of cognitive function. The most flexibility in shifting from one subcategory to
commonly used score from verbal fluency tests another and is a relatively effortful process.
is the total number of words generated. How- These components of fluency performance
ever, this score provides little information about are differentially affected by various neurologi-
the cognitive processes underlying fluency per- cal disorders. Clustering is related to temporal-
formance and does not answer the question as to lobe functioning, as indicated by impaired per-
why a particular patient group or experimental formance among patients with temporal lobec-
manipulation is associated with reduced test per- tomy for intractable epilepsy (Troyer, Mos-
formance. Additional information is needed to covitch, Winocur, Alexander, & Stuss, 1998)
examine the behavioral components that deter- and patients with Alzheimer’s disease (Troyer,
mine fluency performance. Moscovitch, Winocur, Leach, & Freedman,
Optimal fluency performance involves gener- 1998). Clustering is unaffected by focal frontal
ating words within a subcategory and, when a lesions (Troyer, Moscovitch, Winocur, Alexan-
subcategory is exhausted, switching to a new der, et al., 1998). Switching, on the other hand,
subcategory. These behavioral components were is related to frontal functioning. That is, switch-
identified (Bousfield & Sedgewick, 1944; Grue- ing is specifically impaired among patients with
newald & Lockhead, 1980) and operationalized left dorsolateral and superior medial frontal-lobe
as clustering and switching, respectively lesions (Troyer, Moscovitch, Winocur, Alexan-
(Troyer, Moscovitch, & Winocur, 1997). Clus- der, et al., 1998) and is decreased under condi-
tering involves phonemic analysis on phonemic tions of divided attention (Troyer et al., 1997),
fluency and semantic categorization on semantic an experimental model of frontal dysfunction
*
Thanks to Morris Moscovitch and Kathryn Stokes for conceptual input and comments on the manuscript; Jill
B. Rich, Nicole D. Anderson, and Don Stuss for the provision of fluency protocols; Malcolm Binns for statistical
assistance; and Katy Kamkar for assistance with scoring and data entry.
Address correspondence to: A. Troyer, Psychology Department, Baycrest Centre for Geriatric Care, 3560
Bathurst Street, Toronto, ON, M6A 2E1, Canada. E-mail: a.troyer@utoronto.ca.
Accepted for publication: October 19, 1999.
NORMATIVE FLUENCY DATA 371
(Moscovitch, 1994). Decreased switching is also were required either to: (a) obtain scores of 25 or
seen in patient groups with frontal dysfunction higher on the Mini Mental Status Examination
in the context of additional brain dysfunction, (Folstein, Folstein, & McHugh, 1975), or (b) score
within the normal range on an episodic memory
including Parkinson’s disease (Tröster et al.,
test (Anderson, Craik, & Naveh-Benjamin, 1998).
1998; Troyer, Moscovitch, Winocur, Leach, et
al., 1998), Huntington’s disease (Rich, Troyer, Fluency Tasks
Bylsma, & Brandt, 1999), multiple sclerosis All testing was conducted in English. Each partic-
(Tröster et al., 1998), and schizophrenia (Robert ipant was administered three trials of a phonemic
et al., 1998). fluency task and one or two trials of a semantic
Overall, research with these patient groups fluency task. For phonemic fluency, participants
were given either the letters F, A, and S (n = 257)
has consistently shown distinctions between pa- or C, F, and L (n = 154). Consistent with standard
tient groups with predominant temporal versus instructions (e.g., Spreen & Strauss, 1998), partici-
frontal dysfunction. Some inconsistencies have
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
ency, only phonemic clusters were counted, and on sex-specific and age-specific names of the same
semantic fluency, only semantic clusters were animal species were considered to be the same ani-
counted. The size of the cluster was counted begin- mal (e.g., hen and rooster, cat and kitten). In order
ning with the second word in each cluster. The to provide consistency between the semantic flu-
mean cluster size was calculated by summing the ency tasks, on supermarket fluency, no credit was
size of each cluster and dividing by the number of given for subcategory labels (e.g., fruits) if spe-
clusters. Switches were calculated as the number cific exemplars were also given (e.g., apple, ba-
of transitions between clusters, including single nana).
words. The number of switches, therefore, is the On phonemic fluency, scores from the three tri-
same as the number of clusters minus the number als (i.e., F, A, and S, or C, F, and L) were com-
of trials administered (e.g., three trials on phone- bined into a single score for each participant. Simi-
mic fluency and two trials on semantic fluency). larly, on semantic fluency, scores on the two trials
For several reasons, the raw number of switches (i.e., Animals and Supermarket) were combined
was chosen as the switching index rather than the into a single variable. The benefit of these com-
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
number of switches corrected for number of words bined scores is that they are based on a higher
generated. Conceptually, the raw number of swit- number of responses and are thus presumed to be
ches is the behavior of interest, as it indicates the more valid and reliable measures of fluency abil-
number of times an individual can generate a new ity. Because a large number of participants re-
cluster of responses. This is similar to the raw in- ceived animal but not supermarket fluency, animal
dex used for the number of words generated. scores alone were also retained.
Knowing the percent of the words generated that
were correct does not provide as much information
as knowing the raw number of words generated RESULTS
(e.g., 90% could indicate 9 out of 10 or 36 out of
40). As well, because switching is thought to deter-
mine, in part, the number of words generated Raw total mean scores obtained by the entire
(Troyer et al., 1997), correcting switches for total sample were 42.5 (SD = 11.7) for FAS and CFL,
words generated would be tantamount to correct- 19.5 (SD = 5.3) for Animals, and 22.9 (SD =
ing a cause for its effect. Another reason that raw 5.8) for Supermarket fluency. These scores are
switches were used is because previous experience consistent with previously published scores
has indicated that a corrected switching score does
(e.g., Bolla, Lindgren, Bonaccorsy, & Bleecker,
not produce meaningful information. No patient
groups, including those with mild to moderate de- 1990; Kozora & Cullum, 1995; Spreen &
mentia, have been found to be impaired in compar- Strauss, 1998).
ison to controls on a corrected switching score, To determine the contributions of demo-
despite large group impairments on other fluency graphic and administrative factors to fluency
measures (Troyer, 1997; Tröster et al., 1998). performance, a series of regression equations
Importantly, repetitions and intrusions were was performed using age, education, sex, and
included in calculations of cluster size and swit-
ches because they provide information about the
phonemic-fluency test form (i.e., CFL or FAS)
ongoing strategy. Any protocol on which the ex- to predict cluster size, switches, and total words
aminer failed to record repetitions or intrusions in generated on the fluency tasks. Effect sizes (i.e.,
the order in which they were generated was dis- f; Cohen, 1988) based on the partial regression
carded from the analyses (six protocols). This was coefficients (Bs) were calculated to determine
considered to be important because, otherwise, which variables made meaningful contributions
both clustering and switching would be under- to fluency scores. According to Cohen (1988), fs
reported in patient populations in which perse-
verations are frequent. of .10, .25, .40 correspond to small, medium,
Repetitions and intrusions were excluded from and large effect sizes, respectively.
the number of correct words generated. On phone-
mic fluency, consistent with the instructions given Age
to participants, proper names and repetitions of the Age showed a small effect size as a predictor of
same word with a different ending were also ex- switching on phonemic fluency, B = 0.04, t =
cluded. On animal fluency, no credit was given for
–2.05, p = .041, f = .10, and a large effect size as
subcategory labels (e.g., bird) if specific exemplars
were also given (e.g., robin, canary). In addition, a predictor of switching on semantic fluency, B
NORMATIVE FLUENCY DATA 373
increasing age was associated with slightly Generally, age, education, and fluency form (but
larger cluster sizes and with reduced switches not sex) were important predictors of fluency
and words generated. performance. Thus, regression analyses were
performed that included only these three predic-
Education tor variables. Corrections were then determined
Number of years of formal education showed by the B’s, and these corrections were applied to
minimal to small effect sizes as a predictor of the original scores in order to calculate norma-
clustering on phonemic fluency, B = 0.015, t = tive test data. The corrections are presented at
3.34, p = .001, f = .17, semantic fluency, B = the top of Table 1. Age and education correc-
0.012, t = 0.75, p = .454, f = .06, and animal flu- tions were applied for each year of age or formal
ency, B = 0.024, t = 2.31, p = .021, f = .12. Edu- education, respectively, and form corrections
cation had a small effect size as a predictor of were applied for FAS. Percentiles were then cal-
switching on phonemic, B = 0.39, t = 2.53, p = culated based on the actual distribution of the
.012, f = .13, semantic, B = 0.25, t = 1.67, p = corrected scores (e.g., 16% of participants in our
.097, f = .14, and animal fluency, B = 0.17, t = sample obtained scores at the 16th percentile or
3.49, p = .001, f = .17. Education showed the less). This method was chosen because some
largest effect size (small to medium) as a predic- score distributions (i.e., cluster size) were
tor of total number of words generated, includ- slightly skewed. Means, standard deviations,
ing phonemic, B = 1.06, t = 5.06, p < .001, f = and percentiles for the corrected scores are pre-
.25, semantic, B = 0.74, t = 2.80, p = .006, f = sented in the Table.
.23, and animal fluency, B = 0.51, t = 6.12, p < Percentiles for individual raw scores can be
.001, f = .31. Higher levels of education were obtained by adding any relevant corrections and
always associated with better fluency perfor- looking up the corresponding corrected score.
mance. Consider, for example, a hypothetical 50-year-
old woman with 13 years of education. If she
Sex produced a cluster size of 0.35 on FAS, her cor-
Sex showed a minimal effect size as a predictor rected score would be calculated as 0.35 +
of every phonemic, semantic, or animal fluency 50(–.001) + 13 (–0.015) + 0.094 = 0.20. This
variable, t’s = 0.08 to 1.25, p’s = .21 to .94, f’s = places her at approximately the 50th percentile.
.00 to .06, with the exception of a small effect If this same individual generated a total of 35
size on total number of words generated on se- words on semantic fluency, her corrected score
mantic fluency, B = –2.32, t = –1.55, p = .123, f would be 35 + 50(0.23) + 13 (–0.74) = 36.8,
= .13. which places her between the 5th and 16th per-
centiles.
374 ANGELA K. TROYER
Table 1. Corrections, Demographically Corrected Descriptive Data, and Percentiles for Fluency Scores.
Age (years) –0.001 +0.05 +0.04 –0.001 +0.11 +0.23 –0.002 +0.05 +0.09
Education (years) –0.015 –0.38 –1.06 –0.012 –0.25 –0.74 –0.023 –0.17 –0.51
Form (FAS) +0.094 –2.67 –2.18 NA NA NA NA NA NA
Mean 0.24 23.9 28.6 0.94 23.4 46.9 0.75 9.8 18.1
SD 0.23 8.2 11.1 0.47 4.4 7.9 0.57 2.7 4.6
1st percentile –0.16 6.6 4.3 0.24 13.4 28.3 –0.24 3.9 8.3
5th percentile –0.06 10.2 11.4 0.40 16.2 34.4 0.01 5.8 10.9
16th percentile 0.01 15.6 17.0 0.60 18.9 39.4 0.23 7.3 13.5
25th percentile 0.08 18.7 20.6 0.66 20.5 40.7 0.40 7.9 14.9
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
50th percentile 0.19 23.3 28.7 0.91 22.7 46.3 0.64 9.6 17.9
75th percentile 0.35 29.7 36.6 1.18 26.5 52.5 1.12 11.6 21.2
84th percentile 0.44 32.3 39.3 1.44 27.5 56.6 1.39 12.4 22.8
95th percentile 0.73 37.6 47.6 2.02 31.3 60.7 1.89 14.7 26.7
99th percentile 0.97 43.2 57.4 2.37 34.0 62.4 2.43 16.7 29.3
cation was associated with more switches and centile) fluency. This pattern is not consistent
more words generated on animal fluency. In this with Alzheimer’s disease, and instead implicates
case, more frequent switching presumably re- frontal dysfunction, perhaps related to vascular
sulted in more words generated. Thus, both clus- or frontal dementia. Indeed, vascular dementia
tering and switching are necessary to account for was the most likely diagnosis for LS, given her
variations in the number of words generated. history (stable impairment over the last year)
The usefulness of these fluency scores in as- and brain imaging (periventricular white matter
sessment and diagnosis can be demonstrated changes on CT). Thus, an examination of clus-
with the following sample patient profiles. SR is tering and switching scores may be useful in
a 56-year-old woman with a university-level discriminating between different causes of cog-
education and a one- to two-year history of nitive impairment.
memory decline. A comparison of her demo-
graphically-corrected fluency component scores
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
Lacy, M. A., Gore Jr., P. A., Pliskin, N. H., Henry, G. Tröster, A. I., Fields, J. A., Testa, J. A., Paul, R. H.,
K., Heilbronner, R. L., & Hamer, D. P. (1996). Blanco, C. R., Hames, K. A., Salmon, D. P., &
Verbal fluency task equivalence. The Clinical Neu- Beatty, W. W. (1998). Cortical and subcortical in-
ropsychologist, 10, 305-308. fluences on clustering and switching in the perfor-
Mattis, S. (1988). Dementia Rating Scale. Odessa, FL: mance of verbal fluency tasks. Neuropsychologia,
Psychological Assessment Resources. 36, 295-304.
McKhann, G., Drachmann, D., Folstein, M., Katzman, Tröster, A. I., Salmon, D. P., McCullough, D., & But-
R., Price, D., & Stadlan, E. M. (1984). Clinical ters, N. (1989). A comparison of the category flu-
diagnosis of Alzheimer’s disese. Neurology, 34, ency deficits associated with Alzheimer’s and
939-944. Huntington’s disease. Brain and Language, 37,
Moscovitch, M. (1994). Cognitive resources and dual- 500-513.
task interference effects at retrieval in normal peo- Troyer, A. K. (1997). [Alternate indices of verbal flu-
ple: The role of the frontal lobes and medial tem- ency clustering and switching]. Unpublished raw
poral cortex. Neuropsychology, 8, 524-534. data.
Rich, J. B., Troyer, A. K., Bylsma, F. W., & Brandt, J. Troyer, A. K., Moscovitch, M., & Winocur, G.
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
(1999). Longitudinal analysis of phonemic cluster- (1997). Clustering and switching as two compo-
ing and switching during word list generation in nents of verbal fluency: Evidence from younger
Huntington’s disease. Neuropsychology, 13, 525- and older healthy adults. Neuropsychology, 11,
531. 138-146.
Robert, P. H., Lafont, V., Medecin, I., Berthet, L., Troyer, A. K., Moscovitch, M., Winocur, G., Alexan-
Thauby, S., Baudu, C., & Darcourt, G. (1998). der, M. P., & Stuss, D. (1998). Clustering and
Clustering and switching strategies in verbal flu- switching on verbal fluency: The effects of focal
ency tasks: Comparison between schizophrenic and frontal- and temporal-lobe lesions. Neuropsy-
healthy subjects. Journal of the International Neu- chologia, 36, 449-504.
ropsychological Society, 4, 539-546. Troyer, A. K., Moscovitch, M., Winocur, G., Leach,
Spreen, O., & Strauss, E. (1998). A compendium of L., & Freedman, M. (1998). Clustering and switch-
neuropsychological tests: Administration, norms, ing on verbal fluency tests in Alzheimer’s and Par-
and commentary (2nd ed.). New York: Oxford kinson’s disease. Journal of the International Neu-
University Press. ropsychological Society, 4, 137-143.
NORMATIVE FLUENCY DATA 377
APPENDIX
Phonemic fluency
Clusters on phonemic fluency trials consisted of successively generated words which shared any of
the following phonemic characteristics:
First letters: words beginning with same first two letters, such as ‘‘arm’’ and ‘‘art’’
Rhymes: words that rhyme, such as ‘‘sand’’ and ‘‘stand’’
First and last sounds: words differing only by a vowel sound, regardless of the actual spelling,
such as ‘‘sat,’’ ‘‘seat,’’ ‘‘soot,’’ ‘‘sight,’’ and ‘‘sought’’
Homonyms: words with two or more different spellings, such as ‘‘some’’ and ‘‘sum,’’ as indicated
by the participant
Semantic fluency
Clusters on semantic fluency trials consisted of successively generated words belonging to the same
subcategories, as specified below. Commonly generated examples are listed for each subcategory,
although listings are not exhaustive.
Animals
African animals: aardvark, antelope, buffalo, camel, chameleon, cheetah, chimpanzee, cobra,
eland, elephant, gazelle, giraffe, gnu, gorilla, hippopotamus, hyena, impala, jackal, lemur, leopard,
lion, manatee, mongoose, monkey, ostrich, panther, rhinoceros, tiger, wildebeest, warthog, zebra
Australian animals: emu, kangaroo, kiwi, opossum, platypus, Tasmanian devil, wallaby, wombat
Arctic/Far North animals: auk, caribou, musk ox, penguin, polar bear, reindeer, seal
Farm animals: chicken, cow, donkey, ferret, goat, horse, mule, pig, sheep, turkey
North America animals: badger, bear, beaver, bobcat, caribou, chipmunk, cougar, deer, elk, fox,
moose, mountain lion, puma, rabbit, raccoon, skunk, squirrel, wolf
Water animals: alligator, auk, beaver, crocodile, dolphin, fish, frog, lobster, manatee, muskrat,
newt, octopus, otter, oyster, penguin, platypus, salamander, sea lion, seal, shark, toad, turtle, whale
Beasts of burden: camel, donkey, horse, llama, ox
Animals used for their fur: beaver, chinchilla, fox, mink, rabbit
Pets: budgie, canary, cat, dog, gerbil, golden retriever, guinea pig, hamster, parrot, rabbit
Birds: budgie, condor, eagle, finch, kiwi, macaw, parrot, parakeet, pelican, penguin, robin, toucan,
woodpecker
Bovine: bison, buffalo, cow, musk ox, yak
Canine: coyote, dog, fox, hyena, jackal, wolf
Deers: antelope, caribou, eland, elk, gazelle, gnu, impala, moose, reindeer, wildebeest
Feline: bobcat, cat, cheetah, cougar, jaguar, leopard, lion, lynx, mountain lion, ocelot, panther,
puma, tiger
378 ANGELA K. TROYER
Supermarket Items
Downloaded by [University of Western Ontario] at 11:37 11 May 2013
Note. From ‘‘Clustering and switching as two components of verbal fluency: Evidence from younger
and older healthy adults,’’ by A. K. Troyer, M. Moscovitch, & G. Winocur, 1997, Neuropsychology,
11, p. 145-146. Copyright 1997 by the American Psychological Association. Adapted by permission
of the publisher.