384

The Auditory Analysis

Test: An Initial Report*

Jerome Rosner, O.D.

Dorothea P. Simon, MA.
Pittsburgh, Pennsylvania

A new test for auditory perception, the Audi- edged the importance of auditory perception
tory Analysis Test, was given to 284 children in (Evans, 1969;Dykstra, 1968; Smith, 1968), but
kindergarten through grade 6. The instrument, offered little in the way of operational defini-
consisting of 40 items, asks the testee to repeat tions except in such general terms as "listening
a spoken word, then to repeat it again without skills." Many pre-reading auditory training pro-
certain specified phonemic elements - such as a grams have been developed that provide the
beginning, ending or medially-positioned con- child with discrimination tasks that are similar,
sonant sound. Seven categories of item diffi- though not identical, to the test items -
culty were proposed. Test results varied, both apparently with the anticipation that whatever
within and between class groups. Performance basic behaviors are required for competent
tended to improve with age and grade place- auditory discrimination will be realized through
ment Pearson Product-Moment Correlations of repeated practice and generalized to other
individual AAT scores with Stanford Achieve- situations such as reading class. The published
ment Test reading scores yielded significant instruments for testing auditory perception that
relationships (p<01) ranging from .53 (grade
we examined were primarily concerned with
I) to .84 (grade 3). Analysis of errors supported
hearing acuity or, at best, discrimination skills
the validity of test item difficulty and provided
(Weprnan, 1958; Murphy and Durrell, 1949;
direction for the design of a treatment ap-
STAP, 1969). Some provide data supporting
proach to auditory perceptual dysfunction.
the tests' validity for predicting reading per-
The ability to differentiate the sounds of the formance. These tests, however, usually 1
require
language, both in listening and speaking, has responses that are limited to "yes-no, ' "same-
long been recognized as an important factor in different," or the like. Such responses provide a
learning to read. The literature concerned with minimum of information concerning the proc-
the teaching of reading has repeatedly acknowl- esses used to produce the assessed behaviors.

The research reported herein was supported and Few instruments have been constructed that
preprinted by the Learning Research and Developmentattempt to analyze the processes that contri-
Center which is supported in part as a research and bute to auditory perception as related to
development center by funds from the United States reading. One exception, a phoneme blending
Office of Education, Department of Health, Educa- test by Jean Chall (1963), does require behav-
tion, and Welfare. The opinions expressed in this
publication do not necessarily reflect the position or iors more complex than "same-different" dis-
policy of the Office of Education, and no official crimination responses. Chall relates her instru-
endorsement should be inferred. ment to certain reading skills and has provided

40 Journal of Learning Disabilities

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

 385

validation of some significance in a longitudinal The outcome of these investigations made

study with a small group of children. apparent the requirements for the kind of
The Learning Research and Development instrument we were seeking - that is, one that
Center (LRDC) Perceptual Skills Curriculum tests for phonic analysis and synthesis behav-
(Rosner, 1969) recognizes the importance of iors, and requires encoding responses more
auditory perceptual skills in reading. The audi- complex than "yes-no" discriminations. This
tory-motor component of the curriculum is paper reports a test that seems to meet both
based on the rationale that the child's ability to criteria - the child is asked to remember and
differentiate the phonemic elements of the analyze spoken sounds and to demonstrate
spoken language develops as the result of these abilities in his oral response. Descriptions
feedback loops between his production and of the test and method of administration are
hearing of vocal sounds. As the child accumu- presented below.
lates experiences, both his hearing and his vocal The questions asked by this study were: (1)
control gain in the direction of increased Do the test results indicate a range of individual
capacities for discrete functioning. The ability competencies in the behaviors we have defined
to sort out the perceptual elements of verbal as auditory perception? That is, do the children
acoustic information seems vital to the subse- display variations in their ability to replicate
quent skill of reliably ordering these elements spoken sounds, eliminating certain specified
into the symbolic constucts - words - of the phonemic components in their responses? (2)
culture. As the capacity to sort, order, and To what extent do individual test results
synthesize sounds is acquired, refined, and correlate with measures of reading performance
performed more efficiently, the task of reliably at different grade levels? (3) If measures cor-
relating phoneme and grapheme, as required in relate with reading, then what data are provided
learning to read and spell, becomes manageable. by item analyses with respect to the design of
The goals, then, of the auditory-motor curricu- training programs to teach competent auditory
lar component are to insure that each child perceptual skills? (4) What are the implications
acquires the skills needed for competent analy- of these test results to developers of reading
sis and synthesis of the phonemes presented in curricula?
a beginning reading program, and that his
repertoire continues to expand as he progresses SUBJECTS
through that program. The subjects were all of the available children
(N = 284) in kindergarten through grade 6 in a
The development of valid and reliable cri- suburban elementary school situated in Western
terion-referenced curriculum tests for these Pennsylvania. The school population is entirely
behaviors has been one of our major objectives. white and middle class. This school has partici-
During the past two years, we examined a pated in many research projects, and the
number of standardized instruments, including subjects are accustomed to various kinds of
the Rosner Perceptual Survey (Rosner, Rich- testing. The children were uniformly coopera-
man, and Scott, 1969; pp. 40-41) which con- tive and, above the kindergarten level, had no
tains two subtests designed to sample auditory difficulty in understanding the test. Table I
perceptual skills. One is an abbreviated version describes the school population.
of the Slingerland Echolalia test (Slingerland,
1962). In the other, the child is asked to PROCEDURES
reproduce rhythmic clapping patterns (Rosner, The Auditory Analysis Test (AAT) consists of
1966). We gathered data on these two tests and 40 English words varying in length from one to
additional data on two types of discrimination four syllables (see Table II). Each word is
tests that required the child to recognize the pronounced by the examiner who instructs the
presence or absence of an embedded sound in child to repeat the word, then to repeat it again
the Context of a word. Both nonsense words but to omit a specified sound. With three
and common English words were used. exceptions the words were selected on the basis

Volume 4, Number 7, August/September, 1971 41

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

386

that the elimination of a phoneme, phoneme

cluster, or syllable would result in another
English word when pronounced. Since this is an
oral test, we were not concerned with consis-
tency of spoken sound to spelling. The sounds
to be eliminated, as indicated by parentheses on
the test record sheet, include 13 different
consonants, one consonant digraph, two con-
sonant blends, and 12 syllables.
Seven task categories were designated:
I. Omission of the final syllable of a
two-syllable word.
II. Omission of the initial syllable of a
two-syllable word.
III. Omission of the final consonant of a
one-syllable word.
IV. Omission of the initial consonant of a
one-syllable word.
V. Omission of the first consonant of a
consonant blend.
VI.* Omission of a medial consonant.
VII. Omission of a medial syllable.
Their arrangement, as shown above, reflects
our predictions of rank order of difficulty,
starting with the task we assumed to be least
demanding. Based on that order, the test was
constructed with the five easier types in the
first half and the two harder types in the
second half. The words are listed by type in
Table III.
The test items were arranged so that four
consecutive errors in the first half of the test
meant a failure to respond correctly to at least
three different types of words. Four consecu-
tive errors in the second half of the test meant
failure with at least two different types or
words. By ordering the items in this fashion we
sought to avoid collecting distorted data in the
event that we had incorrectly judged the
relative difficulty of the seven types of words.
In addition, this ordering scheme asked for a
different behavior for each item; we thus hoped
to avoid unreliable data due to a response set him to "Say cowboy." After the child re-
or to learning accrued on previous items. sponded, the examiner covered the picture of
The tests were administered individually. To the boy and said, "Now say it again, but
explain the test the examiner first showed the without boy." If the response was correct,
child the top half of a sheet of 8Vi x 11-inch ("cow"), the examiner then exposed the bot-
paper on which pictures of a cow and a boy's tom half of the sheet, showing drawings of a
head had been drawn side by side and asked tooth and a brush, and said, "Say toothbrush,"

42 Journal of Learning Disabilities

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

 387

but without the day," and so on down the list.

The examiner always pronounced the specific
sound (s)(not the letter name) to be omitted. If
the subject had a speech articulation problem,
this was noted by the examiner at the first
repetition of the full test word and was taken
into consideration when the accuracy of the
response in which a portion of the word was
omitted was assessed. If the subject failed to
respond to an item, the examiner repeated it
exactly as he had first said it. If there was still
no response, the examiner recorded a score of
zero and presented the next item. Testing was
discontinued after four consecutive errors.
The test score indicates the total number of
correct responses. Hence, the range of possible
scores extends from 0 to 40. Three examiners
conducted the testing on different children in
kindergarten grade through 2. All children in
grades 3 through 6 were tested by the same
person.
TEST RESULTS
Nine of the SO kindergarten children did not
understand the task — that is, they did not pass
the demonstration items, and they were given
scores of zero. All of the other children passed
the demonstration items (indicating that they
were offered at least four items). Only 18.8
percent of the first grade children responded
accurately enough to be offered all 40 test
items, whereas 79.5 percent of the 6th grade
children remained in the test through all of the
items. Table IV lists the percentage of each
class that had not yet been eliminated from the
test (four consecutive errors) at the 10th, 20th,
30th, and 40th item point. A gap between
kindergarten and grade 1 is clearly indicated.
The other interclass differences anticipate the
changes in the mean scores of the groups as age
and, after the child responded, "Say it again, and grade placement moves upward.
but without tooth." If the child failed either
demonstration item, an attempt was made to Table V summarizes the test scores for the
teach the task by repeating the demonstration seven grade levels. From kindergarten upward,
procedures with the pictures. If he again failed the mean scores increased progressively, with
to make correct responses to both items, testing the major change occurring between kinder-
was discontinued and a score of zero (0) was garten and grade 1. The distribution of scores
recorded. If both responses were correct, the within and between classes is graphed with
examiner withdrew the picture sheet and pro- relative cumulative frequency ogives in Figure
ceeded with the test. He said, "Say birthday," 1. Again, the marked difference between
waited for a response, then, "Now say it again, kindergarten and grade 1 is obvious. In addi-

Volume 4. Number 1, August/September, 1971 43

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

388

FIGURE 1. Relative distribution of AAT ores, grades K - 6.

tion, a merging of the curves for grades 1 and 2 purposes of validation, Auditory Analysis Test
at
^50 causes another unusual gap between the scores were correlated with the sum of the
ogives of grades 2 and 3 in the range of scores stanines on the language arts subtests of the
between P3Q and P70. The graph indicates that Stanford Achievement Test.
over 60 percent of both the first and second
grade children earned scores of 19 or less. In VALIDATION RESULTS
contrast, this score was exceeded by over 67 The answer to the second question raised in
percent of the 3rd grade, 69 percent of the 4th undertaking this study - the nature of the
grade, 80 percent of the 5th grade and 87 correlations between the AAT and academic
percent of the 6th. The first and second grade achievement - is contained, at least in part, in
curves diverge again in the upper ranges. We Table VI. The correlation between AAT and
note that appoximately 1S percent of the first language arts achievement is relatively high,
grade and in excess of 28 percent of the second particularly in grades 2 and 3 where the AAT
achieved scores better than 25, the median of accounts for 39 percent and 70 percent of the
grade 3. The test items seem to impose an variance, respectively, in jcontrast to the 32 and
effective ceiling on the scores of grades 5 and 6. 57 percent that is accounted for by IQ. An
In general, Figure 1 suggests that the demands analysis of partial correlation between AAT and
of the AAT are met more readily by children as language arts, with IQ held constant, shows a
they mature and progress through an academic positive value of .5 or better for grades 2
program. through 5 (p < .01). At grade 6, IQ seems to be
the more decisive factor. Table VII is a sum-
VALIDATION PROCEDURE mation of grades 1 through 6 scatter diagrams
During the month in which this testing was of z-scores for language arts and AAT. We
conducted (April 1970), the children were also calculated z-scores within each class separately
given the Stanford Achievement Test. For from both the language arts and AAT data. The

44 Journal of Learning Disabilities

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

 389

ITHM ANALYSIS
To assess the relative difficulty children had
with individual words in performing the audi-
tory analysis task, item responses were rank
ordered according to several different methods.
Some of the schemes took account of the fact
that not all children were tested on the entire
list; some did not take this into account.
Analysis by all these methods, however, yielded
only small differences in ordering. Further-
more, the first four task "types" (14 words
see Table III) remained at the top of every list.
Table VIII presents the test words ordered by
the simplest, most straight-forward method a
count of correct responses in each class - with
some confidence that this ranking represents
the average difficulty of individual words for
children in grades 1 through 6.
Table IX, which presents the mean percent-
age of correct responses by word type at each
grade level, shows that the children beyond the
first grade found it easier to omit an initial
sound than to omit a final sound. As predicted,
in general it proved harder to omit the first
sound of an initial consonant blend than of an
initial consonant-vowel combination, and still
more difficult to omit a medial consonant. The
longer Type VII words put an additional strain
on short-term memory, and, except for "cre-
ate" and "reproduce," all of these words appear
at the end of the list.
The interclass data indicate that the ability
to respond appropriately to the AAT improved
for all word types in step with maturation
and/or improved reading skills among children
nine cells contain the total number of subjects in kindergarten through 3rd grade. Word Types
in grades 1 through o. Their z-scores have been I through IV were apparently not of sufficient
sorted into one o( three categories: from +1.00 difficulty to provide discriminating powers
upward, between -0.l)l> to +0.1)1), and from within grades 4, 5, and 6. The more difficult
-1.00 downward. Only one child (grade 2) types (V, VI, VII), however, reflect changes
achieved an AAT z-score o\ +1.00 or better and that, again, may be related to age and grade
a language arts <L.A.) z-score- of -1.00 or less. placement.
No child displayed the opposite results that
is, an AAT z-score of -1.00 or less combined DISCUSSION
with an L.A. z-score or +1.00 or better. Thus, The AAT seems to provide a systematic method
in addition to favorable correlation coefficients, of assessing the degree to which a child has
we observe that, in all but one instance, learned to sort, order, and synthesize the
children who performed well above average in perceptual elements of auditory information. In
one of the measured tasks never performed well respect to the initial question asked by this
below average in the other. study, a relatively wide performance range was

Volume 4. Number 7, August/September, 1971 45

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

390

demonstrated within and between the various

age groups; thus, individual variation was
tapped. Are these skills genetically determined?
Are they the result of experience, both inci-
dental and intentional - the latter, perhaps, by
way of reading instruction? The answers, not
fully available in these data, certainly should be
sought. The major differences evidenced be-
tween kindergarten and grade I (acknowledging
that this investigation was conducted near the
end of the school year) may imply that reading
instruction is an important factor in teaching a
child to perform the analytical tasks we meas-
ured. This is a reasonable stance, considering
the general powerful effect that results from
teaching a child to use visual information as a
mediator for auditory. It cannot be concluded,
however, that the AAT is completely depend-
ent upon a child's reading skills or, for that
matter, vice versa. The logical position is that
the two are completely interwoven -- that, as a
child commences to grasp the concepts of
phoneme-grapheme interrelationships, he is
learning the initial decoding skills of reading
which, in turn, teach him to sort verbally
produced perceptual information which, again
in turn, assists him in the auditory sorting and
ordering skills needed for reading, and so on.
More study will be needed to answer this
interwoven puzzle. More importantly, inquiry
must be initiated as to which of the two
behaviors - symbol decoding or auditory analy-
sis may most readily be taught at an early
age. If it is the latter, it makes sense to urge
that this capacity be exploited, albeit benignly,
at least with certain populations.
In regard to the second question, concerning
the relationships of the AAT to reading, here
too the evidence offers reason for further
investigation. The substantial correlation be-
tween the AAT and language arts scores sug-
gests that there may be a cause-and-effect
connection; certainly much more work is
needed, across different populations and in
different settings, but correlational data will
not be sufficient. If, indeed, a true relationship
does exist between the AAT and reading, what
effect will an intervention program that im-
proves AAT performance have upon reading

46

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

 391

TABLE IX. Mean percent correct responses to grade. The prior program was based on teaching
word types by grade level the process of decoding a linguistically regular
sound-symbol system. The new program is
GmdeUvei much more analytical in nature, It initially
Typt K 2 3 4 stresses a precise, structured decoding approach
to reading in which individual letters and
I 80.0 100.0 97.6 100.0 100.0 100.0 100.0 their sounds are taught. After the child has
II 52.0 77.4 82.9 97.3 96.6 100.0 100.0
Ul 20.0 81.8 80.9 91.6 94.9 94.5 94.0 mastered a predetermined number of pho-
IV 7,0 70.2 86.9 93.7 94.8 97.6 97.0 neme-grapheme combinations, he is taught to
V 6.3 44.3 42.9 60.8 63.8 71.1 74.7 blend the individual phonemes into units that
VI 0.5 22.6 33.5 53.1 56.9 62.1 74.1
VII 0.2 13.7 24.6 29J 33.4 38.3 44.9 form the basis of his beginning reading vocabu-
lary. Does the very narrow gap between the
grades 1 and 2 AAT scores, reported here,
achievement? The answer to this question is suggest a better than average first grade, or a
essential before any firm conclusion regarding difference in the teaching method? We do not
cause and effect is to be drawn. yet have any other samples for comparison. If
Yet another factor must be mentioned. In the new teaching system does assist the child in
our judgment, reading is primarily an auditorily developing more analytical auditory skills
based skill. The analysis of the visual symbols which, in turn, support him in learning to read
presented in reading is less demanding, and more efficiently, the AAT may be useful in
perhaps less important to the typical first grade assessing the effectiveness and uniqueness of
child, than the analysis of verbal sounds. Visual that instructional program. If this is the case,
analysis is, nonetheless, a factor in learning to there are implications for pre-reading instruc-
read. With this in mind, we examined the effect tion in the skills needed for successful AAT
upon the correlation between L.A. and AAT at performance.
the first grade level after a visual-motor score*
was added to the AAT. The correlation coeffi- One final consideration: Are there individual
cient did rise - from .53 to .62, accounting for differences in modality preference? Are there
an additional ten percent of the variance. The certain children who would profit more from a
correlation between the first grade visual-motor program that strongly stressed the visual differ-
scores and language arts, omitting the AAT ences of graphemes before introducing the
scores, on the other hand, was only .32. phonetic relationships? Are there children of
The third question asked what could be the opposite type - those who initially should
learned from the item analysis regarding the be taught to analyze verbal sounds and, once
structuring of a training program. Enough data having achieved some skills in sorting and
are available to support the basic sequence of ordering the acoustic elements of the language,
the word types. A thorough analysis of error be taught the graphic code by relating it to
responses directs us to a more reliable and valid sound? Is it logical to expect children to
AAT, and, what is even more important to a progress equally well with either of the two
refined recognition of the criteria for struc- approached or some combination of them? Not
turing the learning objectives into a hierarchy. necessarily; yet in many schools the same
The fourth question, the implication of this reading program is provided for all. Is it
type of study to reading curriculum designers, coincidental that in many schools, also, a
is an important and interesting one. The school significant percentage of the students display
in which our testing was conducted introduced reading problems not explainable by IQ or
a new early reading program to this year's first other predictors? Clearly, the answers are not
yet available; and indeed these questions merit
*The visual-motor score is derived from a quantitative
investigation. Individual differences - apti-
assessment of the subject's responses to the Gesell tudes - do exist. Consideration should be given
Copy Forms and Rutgers Drawing Test. to the design of instructional programs that

Volume 4, Number 7, August /September. 1971 47

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015

392

acknowledge individual differences in percep- Murphy-Durrell Diagnostic Reading Readiness Test.

tual aptitudes, identify them, and teach to the New York: Harcourt,Braced World, 1949.
Rosner, J: Visual Perceptual Survey, An evaluation
student's weaknesses through his strengths.
designed for the Oakmont Public School District,
The AAT, in our judgment, provides an Oakmont, Pa., ESEA Title 1, 1966. Mimiographed
additional tool for recognizing those individual report,
differences. Rosner, J.: The design of an individualized perceptual
skills curriculum. LRDC Working Paper #5J.
University of Pittsburgh Learning Research
Pittsburgh: Univ. of Pittsburgh, 1969.
and Development Center Rosner, J., Richman, V.t and Scott, R, H.: The
160 North Craig Street (15213) identification of children with perceptual-motor
dysfunction. LRDC Working Paper #47. Pitts-
Referencet
Chall, J.: A uditory blending ability: a factor in success burgh: Univ. of Pittsburgh, 1969.
in beginning reading. Reading Teacher, 1963, 17,Slingerland, B.: Screening test for identifying children
113-118. (See also Roswell-Chall Auditory Blend- with specific language disabilities. Cambridge,
ing Test. New York: Euay Press, 1963). Mass: Educators PubL Service, 1962.
Dykstra, R.: Continuation of the coordinating center Smith, H. K. (Ed.): Perception and Reading. Proceed-
for first grade reading instruction programs. Report ings of the Twelfth Annual Convention of the
of Project N. 6-165L Minneapolis: Univ. of Minn., International Reading Association, Vol. 12, Part 4.
1967. (Summary of report in Reading Research Newark, Del: Internet. Reading Assn. 1968.
Quart.,1968, 4, 49-70). The STAP: Screening Test for Auditory Perception:
Evans, J. R.: Auditory and auditory-visual integration San Rafael, Calif: Academic Therapy Pub I. 1969.
skills as they relate to reading. Reading Teacher, Wepman Auditory Discrimination Test. Chicago: Wep-
1969, 22, 625-629. man, 1958.

NEWS
THE CHILDREN'S TELEVISION Workshop's incorporating the show into regular classroom
new TV show for teaching reading skills to instruction is being prepared for teachers,
children between the ages of seven and 10 will administrators and curriculum supervisors.
be on the air in September. It was described by The CTW Director of Research Edward L.
several Workshop officials at the annual meet- Palmer indicated that pretesting and posttesting
ing of the International Reading Association. of the series will be conducted by both his staff
The new show will have an identity separated and an independent evaluation group to meas-
from Sesame Street, acknowledged Samuel V. ure the impact of the series. He also revealed
Gibbon, who heads the reading effort. It will be that five full-length prototype shows are being
broadcast in the late afternoon and will also be presented to a selected target audience, and a
available earlier in the day for in-school use. test battery is being developed by the Center for
Each program will be independent, so that the the Study of Evaluation at UCLA to determine
viewer would not be required to have seen any curriculum objectives.
previous program. "Within each program," Gib-
EDWARD C. FRIERSON, director of the Nash-
bon asserted, "the instruction would be se-
ville Learning Center, author and editor, will
quenced from simple to complex, reflecting the
speak at the fall symposium of the Association
entire range of skills represented in the curric-
of Learning Disabilities Teacher-Consultants,
ulum."
October 2, 1971. The symposium is scheduled
Ahwanetta Cutler, the curriculum consultant at the East Brunswick High School, East Bruns-
for the show, said that several curriculum wick, N. J., from 9 a.m. to 1 p.m., and costs $1
principles and strategies would be utilized, and for members, $5 for non-members. Register
printed materials will be distributed to both through Noreen Gallagher, Morris County Dept.
children and teachers. Also, a guide including of Education, Court House, Morristown, N. J.
curriculum descriptions and suggestions for 07960.

48 Journal of Learning Disabilities

Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on January 9, 2015