Content uploaded by Jordan Richard Schoenherr
Author content
All content in this area was uploaded by Jordan Richard Schoenherr
Content may be subject to copyright.
Category Properties and the Category-Order Effect
Jordan Schoenherr (psychophysics.lab@gmail.com)
Department of Psychology, Carleton University
1125 Colonel By Drive, Ottawa, ON K1S5B6 Canada
Robert Thomson (rthomson@connect.carleton.ca)
Department of Cognitive Science, Carleton University
1125 Colonel By Drive, Ottawa, ON K1S5B6 Canada
Abstract
Previous research has demonstrated that recall performance is
facilitated by the order of presentation of salient stimuli,
referred to as the category-order effect (Brooks & Watkins,
1990; Greene & Lasek, 1994). Specifically, recall is higher on
the full list when numbers precede words. To extend and
clarify this research, Experiment 1 modified Brooks and
Watkins’ paradigm by presenting items as a unitary string (as
opposed to a sequential list) containing four random non-
repeating numbers and four letters representing four
conditions: random letters, rhyming letters, four letter pseudo-
words (consonant-vowel-consonant-vowel strings), and four
letter words. In contrast to past results, this experiment found
better list recall when words and pseudo-words preceded
numbers. Experiment 2 more closely represented Brooks and
Watkins’ (1990) paradigm by sequentially presenting items in
2 blocks: a block of four numbers and a block of four letters.
Keywords: category-order effect; categorization; list recall;
memory; lexicality; working memory
Introduction
Two of the most fundamental aspects to memory are the
items stored within it and the order in which these items are
encoded and retrieved (Healy, 1974; Lashley, 1951).
Although there has been considerable debate over the
degree to which this information is inseparable or whether
they are in fact independent (Murdock & Von Saal, 1967),
there is definitive evidence that the ability to group
information together facilitates recall (Ryan, 1969).
The present study examines the conjoint influence of item
and order information in terms of the category-order effect
(Brooks & Watkins, 1990).
The Category-Order Effect
The category-order effect refers to improved performance
on list recall when a set of items from a small, homogeneous
category are presented prior to those from a comparatively
larger, heterogeneous category (Greene & Lasek, 1994).
Category-order effects have been established across several
word categories, including when high-frequency words
precede average-frequency words (Watkins & Watkins,
1977), semantically-related words precede unrelated words,
rhyming words precede non-rhyming words (Brooks &
Watkins, 1990), and when numeric digits precede words
(Greene & Lasek, 1994; Brooks & Watkins, 1990).
However, the category-order effect can be eliminated by
decreasing the presentation rate of the stimuli or by using
articulatory suppression (Greene & Lasek, 1994).
Greene and Lasek (1994) explored whether the order of
recall of the stimuli (i.e. input or output position) was a
factor in the category-order effect. To test their hypothesis,
participants were asked to either recall items in forward or
backward order. In the backward condition, items were to be
recalled in reverse order that they were presented in. An
additional backward recall condition was also conducted
that had participants recall the categories in reverse order,
but the items within them in a forward order. Although
recall order did affect recall accuracy and overall memory
span, both forward and backward recall exhibited a
significant category-order effect, with improved recall when
more readily categorized items were presented in the first
half of the list. Thus, they concluded that input position was
the primary determinant of the category-order effect.
Enhancing Memory Capacity
The improved performance on list recall evidenced in the
category-order effect is consistent with several results from
the broader field of categorization. Landauer and Freeman
(1968) demonstrated that the size of a category is inversely
related to recognition time. Within categories, common
features facilitate recognition of similar members (Feldman,
2003). In terms of lexicality, familiar (i.e. salient) words are
more accurately recalled from long-term memory (Waugh &
Norman, 1965). Lastly, it has also been shown that digit
span is greater than word span (Miller, 1956).
A notable contrasting result is the performance gain
exhibited by rhyming words in Brooks and Watkins (1990).
Elements that rhyme are generally not as well remembered
when presented in earlier portions of a to-be-remembered
list, whereas those that are subsequently presented have
been known to facilitate list recall (Baddeley, 1986).
Additionally, although the category-order effect seems a
natural extension of what is already understood about
working memory processes, there have been numerous
issues in replicating Watkins’ (1977) original findings. As
Brooks and Watkins (1990) note, the failure of Klauser and
Puckett (1979) is particularly troublesome given that, using
Watkins’ (1977) stimuli and methodology, no category-
order effect was evident in either university students or
elderly participants.
Serial Position and the Word-Length Effect
A limitation of category-order studies is the neglect of serial
position analyses, especially given that serial position
effects are directly concerned with the recall order of
subspan lists and have been robustly documented in the
memory literature. Two principle findings have been the
facilitation for recall of items occupying initial and terminal
positions of a list: i.e., primacy and recency effects. These
position effects are further altered by the order in which
items are recalled. Forward recall is characterized by large
primacy and small recency effects, while the reverse trend is
observable for backward recall with larger recency effects
compared to primacy effects (Atkinson & Shiffrin, 1968).
These findings have also been replicated within probe-
recognition paradigms (Waugh & Norman, 1965).
Another potential confound within this literature is a lack
of control for word length. The word-length effect is the
phenomenon where words of relatively longer length have
reduced recall rates when compared with words of shorter
length (Baddeley, 1986). The word stimuli used in Brooks
and Watkins (1990) are related across semantic categories,
but vary in their length.
Although it is difficult to imagine creating lists
constituted of words of equal length and strongly related, it
represents a potential concern given that both Brooks and
Watkins’ (1990) and Greene and Lasek’s (1994) analyses
are performed on aggregate scores and do not examine serial
position or control for word-length effects.
Lexicality and the Category-Order Effect
A fundamental and transparent finding is that, in memory
tasks, words are extremely salient stimuli. Reichler (1969)
first demonstrated this in terms of a word superiority effect,
where participants were more accurately in identifying a
critical letter in the recall phase when that letter was
embedded in a word rather than a non-word. This effect
potentially arises from a greater number of common features
being identified from long-term memory with word stimuli
than with non-word stimuli. Alternatively, it may be that it
is perceptually simpler to interpret information within word
stimuli.
One method of differentiating these hypotheses is through
the examination of pseudo-words: stimuli with word-like
phonotactic properties but are not part of a subject’s lexicon.
These words are generally constructed in a consonant-
vowel-consonant sequence. Pseudo-words enjoy similar
articulation rates to words but have reduced recall when
compared with words (Hulme, Maughan, & Brown, 1991).
Present Research
The present study examines several category properties to
better understand the mechanisms behind Brooks and
Watkins’ (1990) and Greene and Lasek’s (1994) results that
numeric digits preceding words facilitates full-list recall. A
corollary of this result is that, at least under their task
demands, numeric digits are more salient than words.
Experiment 1 modifies Brooks and Watkins’ (1990)
methodology by dividing stimuli into four letter categories:
words, pseudo-words, rhyming letters, and random letters.
These stimuli were paired with randomly organized numeric
stimuli and were presented as a unit. Word length was
controlled and results were scored by position to determine
serial position effects as a possible explanadum. Experiment
2 more closely replicated Brooks and Watkins’ (1990)
original methodology by presenting each stimulus category
in a unitary fashion.
Experiment 1
The authors reasoned that a letter sequence representing a
word or pseudo-word would be more readily categorized
than a sequence of non-repeating numbers, especially if
those numbers had no extrinsic meaning such as a specific
date or time. Numeric stimuli will still have some salience
when presented as a random sequence, as we observe them
on a daily basis in the form of monetary values, dates, and
other metrics. Under these same conditions, random letters
should not exhibit comparable facilitation provided they are
not organized into meaningful combinations such as
acronyms.
If the category-order effect is simply a function of items
sharing membership in the same homogenous category, then
numbers and rhyming letters should display significant
category-order effects given that they represent a smaller
subset of potential digit combinations. Alternatively, if the
basis of the category-order effect is indicative of higher-
order grouping, such as in lexical categories, category-order
effects should arise solely where lexical properties are
exhibited. This would result in both word and pseudo-word
conditions exhibiting relatively higher recall, with words
having the most robust results. These results will not
necessarily suggest that the category-order effect is limited
to lexical and phonotactic properties, simply that it is
indicative of a case where a hierarchy of categories
improves chunking strategies.
Method
Participants
Twenty-two undergraduates (15 females, 7 males)
participated, and were awarded 1% toward their final grade
in an introductory psychology class.
Materials
Each stimulus consisted of eight items divided into two
homogenous halves belonging to either the number or letter
categories (e.g. BONE1234). Numbers between 1 and 9
were used in a string of 4 non-repeating digits. Numbers
expressing historical dates were excluded.
The same method was used to generate the random letter
and rhyming letter conditions. A subset of rhyming letters
was gathered from the Handbook of the International
Phonetic Association (1999). Again, any combinations that
resembled words or pseudo-words were excluded.
For pseudo-words, consonant-vowel-consonant strings
(CVCs) were gathered from Hilgard (1951) and a random
non-rhyming vowel was appended to the stimuli. The
modifications were conducted bearing in mind the rules for
creation of such stimuli suggested by Luh (1922; cited by
Hilgard, 1951). The Hilgard CVCs used as base stimuli
have a 53% association with real English words. All
pseudo-words were checked against the Merriam-Webster
Dictionary (2002) to ensure that no such words exist. Five
stimuli were replaced as a result of this verification
procedure.
Four-letter words were randomly selected from the
Merriam-Webster Dictionary. Those words that had
repeated letters were excluded from the set. In total, 160
unique experimental stimuli were developed. An additional
20 were also created for the purposes of training, 5 for each
letter condition.
Apparatus
The experiment was designed and presented with Cedrus
SuperLab Pro Version 2.0 on a 17" monitor operating at a
resolution of 1024x768 pixels at 85Hz. As this experimental
software package limited the response input to one
character, an answer sheet was developed with 160 response
fields, each with 8 positions where the participants were to
identify the appropriate stimuli.
Procedure
Participants were tested individually in a dimly lit room.
Participants were told that an eight-item sequence of four
letters and four numbers would be presented, with the
numbers random and the four letters conditions. Each
participant was provided with an answer sheet and
instructed to write down the memory stimuli after a
response cue indicating the direction of recall, either
“FORWARD’ or “BACKWARD”. The response cue
followed a 250 ms inter-stimulus interval that occurred after
the stimuli were presented. If the cue indicated forward,
participants would write down the stimulus in the format it
was presented in. Alternatively, if the cue indicated
backward, participants would be required to respond with
the order of the categories switched while preserving the
order of the items within the category. Instructions
emphasized speed and accuracy equally.
Twenty training trials were presented prior to the
experimental blocks. The stimuli were presented in an
identical manner to the experimental stimuli. This was done
to ensure that the participants were familiar with the
procedures and the type of stimuli being used.
The participants then performed two experimental blocks
of trials, each of which used all 160 unique stimuli.
Participants were randomly assigned to a distractor or no-
distractor condition first, after which they received the
alternate condition. During one block, a ‘+’ would appear in
a random quadrant 250 ms before the memory stimulus as a
proactive attention distractor. The same stimuli were
presented during distractor and no distractor blocks with the
opposite recall cue used in one block relative to the other.
Stimuli were presented for 750 ms. Trials would occur
randomly in accord with SuperLab 2.0 randomization
controls. This resulted in 320 experimental trials, evenly
distributed between the word categories.
Scoring
Participants were scored according to an all-or-none scoring
of individual items when a correct item appeared in the
correct order. In order to compare our current experiment
with those of previous studies (e.g., Greene & Lasek, 1994)
we first aggregated the score for each position. If a correct
item appeared in the correct position as the memory
stimulus, it was awarded 1.0. In the aggregated position
analyses, all positions were summed together for a
maximum of 8.0 points.
Results
Aggregated Position Analyses
A repeated-measures ANOVA was conducted using
accuracy as the dependent variable. The factors that were
examined included letter category (random letters, rhyming
letters, pseudo-words, and words), category order (letters
first or second), recall order (forward or backward), and
score component (numbers or letters). Although the
analysis treats the data as though there are only 4 categories
constituted of letters, it is beneficial to observe that there are
five categories with the inclusion of numeric stimuli.
However, since the numeric stimuli are present in every
display, they are treated as a separate variable. In addition,
the distractor was also used as a within-subjects variable to
replicate Greene and Lasek’s (1994) Experiment 3.
Our data was first analyzed as an aggregate score, which
did not consider either the individual positions or score
component. This was done in order to create a direct
comparison between the present study and previous
research. The following analyses present results in
Greenhouse-Geisser adjusted values. Only those results
involving the main effects or interactions involving
category-order are reported.
In the present analyses, the main effect of category order
demonstrated significance, F(1, 22) = 22.828, MSE = .317,
p < .001, η2
p = .509. The mean recall score demonstrated an
improvement in performance when letters preceded
numbers, rather than the reverse order. This main effect was
qualified by the interaction category order and letter
category, F(3, 66) = 12.806, MSE = .230, p < .001, η2
p =
.368. In general, pseudo-words or words were recalled with
greater accuracy compared to random or rhyming
tetragrams, but this difference was substantially greater
when they preceded number tetragrams compared to when
they were presented afterward. An additional interaction of
category order and score component was also significant
F(1, 22) = 8.570, MSE = 13.702, p < .01, η2
p = .280. This
interaction demonstrates that recall performance was
substantially greater for letter stimuli when presented first
(M = 3.22), then when presented second (M = 2.52) but that
recall performance for numeric remained relatively constant
over this manipulation (M = 2.0 and M = 2.43,
respectively). Lastly, the above interactions were qualified
by a significant 3-way interaction between category order,
letter tetragram and score component, F(3, 66) = 7.728,
MSE = 2.535, p < .005, η2
p = .260 . This finding
demonstrates that although the category order effect is
essentially additive when examining recall performance for
letters, when we examine recall on the numeric tetragrams,
performance increases only when they preceded random or
rhyming letters. Lastly, although distractor did not have an
overall effect on recall, it did significantly interact with
category order and score component, F(1, 22) = 4.549, MSE
= 1.010, p <.05, η2
p = .171. The performance gains
constituting the category-order effect were somewhat
reduced when a distractor was presented before the stimuli.
This seems to indicate reduced attention for the tetragram in
the first position due to resources allocated to the distractor.
Serial-Order Analyses
A second analysis was performed on the same data with two
important differences. First, numeric stimuli were included
with letter stimuli to examine the effect of each category of
tetragram (random letters, rhyming letters, pseudo-words,
words and numbers). Second, each position was coded
separately to obtain a fine-grained analysis of category-
order effects on the positions. This was done to examine
whether some or all of the positions in a display enjoy
performance improvements.
In that it was based on the same data as the analysis
reported above, it has many of the same features. After
reviewing the similarities we will then examine the
differences in these results.
The category-order effect was again significant but was
subject to a decrease in effect size in comparison to the
initial analysis, F(1, 22) = 19.195, MSE = .116, p < .001, η2
p
= .466. Category-order also interacted with tetragram
category, F(4, 88) = 8.588, MSE = .122, p < .001, η2
p =
.281. As the mean recall performance in Figure 1
demonstrates, both random and rhyming letters as well as
numeric tetragrams were unaffected by the order of category
presentation, whereas pseudo-word and words exhibited
reduced accuracy when presented second. Error bars
represent Bonferroni significance values.
When position was included in the analysis it proved to be
significant, F(3, 66) = 66.113, MSE = .088, p < .001, η2
p =
.750. In line with the literature of serial-position effect, item
recall exhibited both primacy and recency effects. This
effect was qualified by an interaction with category-order,
F(3, 66) = 4.842, MSE = .023, p < .05, η2
p = .180. When
letter stimuli were presented first there was an overall
increase in performance relative to when numbers were
presented first. Additionally, when number tetragrams were
presented first recency effects decreased making the final
positions almost equal.
EFFECT OF CATEGORY-ORDER
OVER TETRGRAM CATEGORY
TETRAGRAM CATEGORY
RANDOM RHYMING NUMBER PSEUDO-WORDS WORD
P(COR)
0.4
0.6
0.8
1.0
CATEGORY PRESENTED FIRST
CATEGORY PRESENTED SECOND
In this analysis, distractor did not have any significant
effects alone or interact with category order.
Discussion
This experiment demonstrated that recall performance is
improved when more salient stimuli precede less salient
stimuli. Unlike previous studies, it shows that numeric
stimuli do not enjoy primacy over all stimuli. When placed
on a par with randomly generated letter stimuli, numbers did
demonstrate relatively higher recall.
More generally, the results of Experiment 1 indicate that
the category-order effect is the result of a complex interplay
of cognitive mechanisms. The distractor manipulation only
interfered when we considered each score component.
When examining over position, the effect of distractor
resulted in decreased performance when it was presented
prior to a salient stimulus.
Experiment 2
Experiment 2 was designed to offer a further level of
commensurability between past findings of the category-
order effect. Since this was not a full replication of either
Brooks & Watkins (1990) or Greene & Lasek (1994) it may
be that some elements are missing which lead to their
results. One element that was conspicuously absent in
Experiment 1 was successive presentation of stimuli. It may
be that unitary presentation causes greater primacy effects
while temporally sequential presentation leads to greater
recency effects. To this end, stimuli were broken up into
their constituent score components, which were presented
one after another. Presentation rate was also sped up to limit
the encoding advantage of having fewer digits
simultaneously presented.
Method
Participants
Twenty-two undergraduates (13 females, 6 males)
participated, and were awarded 1% toward their final grade
in an introductory psychology class.
Materials and Procedures
The present experiment segmented the identical stimuli
from Experiment 1 by score component, and presented them
in rapid succession (i.e. present BONE then 1234) at a rate
of 250ms/score component. Procedures and apparatus were
the same as in Experiment 1.
EFFECT OF INPUT AND OUTPUT POSITION ON TETRAG RAM RECALL
LETTERS FIRST
FORWARD BACKWARD
P(COR)
0.0
0.2
0.4
0.6
0.8
1.0
RANDOM LETTERS
RHYMING LETTERS
PSEUDO-WORD
WORD
NUMBER
NUMBERS FIRST
FORWARD BACKWARD
P(COR)
0.0
0.2
0.4
0.6
0.8
1.0
Results
Serial-Order Analyses
An ANOVA with the same dependent and independent
variables was used in the current experiment as in
Experiment 1. Instead of an aggregate analysis, the present
experiment confined itself to an examination over serial-
order.
Replicating the serial-order analysis of experiment 1, the
effect of category-order was significant, F(1, 18) = 6.137,
MSE = .038, p < .05, η2
p = .254. When letters preceded
numbers recall improved (M = 0.54) compared to when
numbers preceded letters (M = 0.52). The interaction of
category-order and recall order was also significant, F(1,
18) = 13.441, MSE = .019, p < .005, η2
p = .427. Participants’
average recall was uniformly higher when they were
required to report letters first followed by number stimuli.
This effect was significant larger when words were
presented first than when they were presented second. The
interaction of category-order and tetragram category was
also significant, F(4, 72) = 4.529, MSE = .047, p < .01, η2
p =
.201. Interestingly, recall for numeric tetragrams was
uniformly greater than any of the letter tetragrams. When
rhyming letters or pseudo-words were presented first, they
were recalled with greater accuracy than rhyming letters or
words. When number stimuli were presented first, letter
stimuli were recalled at an equivalent rate. Qualifying all of
these interactions was the significant three-way interaction
of category-order, recall order and tetragram, F(4, 72) =
8.526, MSE = .065, p < .005, η2
p = .321. As Figure 3
demonstrates, a category-order effect was present for
numbers when they were presented first but recalled second,
for random and rhyming letters as well as pseudo-words
when presented second but recalled first and for words when
presented first and recalled second. As in experiment 1,
error bars represent Bonferroni significance values. The
result again demonstrates the robustness of the category-
order effect while also providing further evidence for
interference that can result from rapid presentation.
Discussion
Experiment 2 demonstrated improved recall performance
when letters were presented first even under conditions that
otherwise favoured the recall of number tetragrams. When
stimuli were presented quickly, recall for numeric stimuli
was significantly greater than all letter stimuli. However,
when there was a category-order effect it favoured recall
performance for letter stimuli.
Interestingly the category-order effect for letter tetragrams
occurred within different recall orders, indicating that the
use of item information varies depending on the temporal
order that a stimulus is presented and retrieved in. The rapid
presentation rate is no doubt the driving factor behind the
reduced recall for word stimuli, but this effect needs to be
further qualified as it only occurs when letters are presented
and recalled in the second position.
General Discussion
The present research replicated a category-order effect
similar to that observed by Brooks and Watkins (1990) and
Greene and Lasek (1994). Previous research demonstrated
that when numbers preceded words, recall performance on
the entire list was improved. In contrast, in Experiment 1
we found that while a category-order effect is indeed
evident within the overall score, it only exhibits improved
performance for word and word-like stimuli, with no score
improvement with random letters, rhyming letters, or
random numbers. By increasing stimulus presentation rate
and mode of presentation in Experiment 2, we again
observed a category-order effect but one in which rhyming
letters now exhibited the effect under certain conditions.
Rapid presentation also nearly eliminated the category-order
effect in word stimuli. This supports early work that
presentation rate can nearly erase the category-order effect
(Greene & Lasek, 1994).
The difference between the present studies and other
research on the category-order effect are instructive. In
Experiment 1, each stimulus was presented as a single
string, which would most likely facilitate grouping of items
in memory at the time of encoding. This should
consequently reduce demands on the working memory
system resulting in improved performance for the overall
scores. A final contribution of Experiment 1 was evident
when position was included as a dependent variable;
demonstrating that a serial-order analysis is an effective
means to study the category-order effect.
In Experiment 2, the letter and number tetragrams were
presented in rapid succession instead of a single string.
Consequently, participants have little time to examine the
stimuli to encode properties that could be used to access
comparable information in long-term memory. Even though
we did not observe a strong category-order effect for
number tetragrams, it appears that successive presentation
favours encoding and retrieval of numeric information when
it is placed in competition with letter information. This
performance gain is offset by improvements resulting from
the category-order effect for letter stimuli
Further study is required to bridge the gap between the
present study and past research on the category-order effect.
Taken together with other studies, the present experiments
suggest four key findings. First, the category-order effect is
a robust phenomena that is evident in a number of stimuli
sets over diverse methods of presentation. Second, this
effect exists at both a macro-level (aggregate scores for
lists) and a micro-level (positional effects). Third, the
category-order effect is relative to the two categories under
consideration. Last, and perhaps most important, the
category-order effect is not limited to instances when a
small homogenous category is presented before a larger
heterogeneous category. Instead, knowledge stored in long-
term memory and the presentation mode can facilitate the
encoding and retrieval of information in working memory.
One limitation of this study should be noted. This
research only controlled for the lexicality of the letter
stimuli while preventing the number stimuli from being
facilitated by any extrinsic knowledge or intrinsic features.
A possible direction for future research would be to use
historical dates (e.g., 1939), numbers with regular increases
in magnitude (e.g., 1234 or 3579), or the binary sets /
repeating sets (e.g., 1011).
In short, our findings suggest that the category-order
effect is a result of a complex interaction of numerous well-
established properties of memory. When stimuli can be
grouped together on the basis of categorical properties (e.g.,
lexicality) stored in long-term memory, recall on a list is
improved when they are presented prior to stimuli from a
category with knowledge that is less accessible. In this
sense, the category-order effect can be understood as a
significant measure of the interaction of cognitive processes.
The present study highlights the need to analysis these
phenomena at an appropriate level to ensure that the
contributions of other memory process are taken into
account. Acknowledgements
The authors would like to thank Sylvain Pronovost and Michael
Henighan for their aid in coding the experimental results and past
collaborative efforts.
References
Atkinson, R.C. & Shiffrin, R.M. (1968). Human memory: A
proposed system and its control processes. In K. W. Spence &
J.T. Spence (Eds.), The Psychology of Learning and Motivation
(Vol. 2). London: Academic Press.
Baddeley, A. D. (1986). Working Memory. Oxford: Clarendon
Press.
Brooks, J. O., & Watkins, M. J. (1990). Further evidence of the
intricacy of Memory Span. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 16, 1134-1141.
Feldman, J. (2003). The simplicity principle in human concept
learning. Current Directions in Psychological Science, 12(6),
227-232.
Greene & Lasek (1994). Category-order effects in memory span.
Journal of Experimental Psychology: Learning, Memory, and
Cognition, 20, 1391-1395.
Healy, A. F. (1974). Separating item from order information in
short-term memory. Journal of Verbal Learning and Verbal
Behaviour, 13, 644-645.
Hilgard, E. R. (1951). Methods and procedures in the study of
learning. In S. S. Stevens, Handbook of experimental
psychology. New York: Wiley.
Hulme, C., Maughan, S., & Brown, G. (1991). Memory of
familiar and unfamiliar words: Evidence for a long-term
memory contribution to short-term memory span. Journal of
Memory and Language, 30, 685-701.
International Phonetic Association (1999). Handbook of the
International Phonetic Association: A guide to the use of the
International Phonetic Alphabet. England: Cambridge
University Press.
Kausler, D. H., & Puckett, J. M. (1979). Effects of word frequency
on adult age differences in word memory span. Experimental
Aging Research, 5, 161-169.
Landauer, T. K., & Freedman, J. L. (1968). Information retrieval
from long-term memory: category size and recognition time.
Journal of Verbal Leaning Behaviour, 7, 291-295.
Lashley, K. S. (1951). The problem of serial order in behavior. In
L. A. Jeffress (Ed.), Cerebral mechanisms in behavior. The
Hixon symposium (pp. 112–136). New York: Wiley.
Miller, G. A. (1956/1994). The magical number seven, plus
or minus two: some limits on our capacity for processing
information. Psychological Review, 101, 343–352.
Murdock, B. B., Jr. and von Saal, W. (1967). Transpositions
in short term memory. Journal of Experimental
Psychology, 74, 137-143.
Ryan, J. (1969). Grouping and short-term memory: Different
means and patterns of grouping. Quarterly Journal of
Experimental Psychology, 21, 137-147.
Reichler, G. M. (1969). Perceptual recognition as a function of
meaningfulness of stimulus material. Journal of Experimental
Psychology, 81, 275-280.
Watkins, M. J. (1977). The intricacy of memory span. Memory &
Cognition, 5, 529-534.
Waugh, N. C., & Norman, D. A. (1965). Primary Memory.
Psychological Review, 72, 89-104.