EPIDEMIOLOGIC STUDY: SUBJECTS AND SAMPLING
target for this study was a stratified cluster sample of 6,000 kindergarten
children who were monolingual speakers. The sample was stratified by residential
setting and clustered according to school building. Rather than choosing a single
urban, suburban and rural area, the sample was drawn from various regions of
the states of Iowa and Illinois. These regions were centered on large metropolitan
areas that will hereafter be referred to as "population centers."
Each population center was selected for the ability to contribute an urban sample,
with the surrounding areas contributing the suburban and rural samples. Several
population centers reduced potential bias in participant characteristics associated
with a single geographic area. The four selected population centers were Des
Moines, Cedar Rapids, Waterloo/Cedar Falls, and the "Quad Cities"
that straddle the Mississippi River. The Quad Cities are Davenport, IA, Bettendorf,
IA, Moline, IL and Rock Island, IL.
Iowa is considered overall to be a rural, farming state, the use of population
centers provided the desired urban, suburban and rural residential strata. The
selected population centers were the largest in the state. Des Moines, the capital
and largest city, had a 1990 city population of 191,000 and a metropolitan area
population of 338,000. The second largest city in Iowa is Cedar Rapids, with
a 1990 city population of 110,000 and a metropolitan area population of 170,000.
Davenport had a 1990 city population of 103,000 and a metropolitan area population
of 384,000. Waterloo/Cedar Falls together had a population of 100,388.
summary, the population centers selected provided a suitable sample for the
study of SLI in monolingual English speaking children. The general population
in the areas sampled provided the linguistic homogeneity desired to reduce the
chances that the identified language deficits were confused with cultural and
The targeted 6,000 kindergarten children were equally distributed into three residential strata: urban, suburban and rural settings. This stratified sampling was specified by the NIDCD contract, and allowed the sampling of children across a spectrum of living and demographic conditions. To achieve this stratified sampling, the attendance zones of the school buildings from the four population centers were drawn and designated as being predominately urban, suburban or rural. Subsequent to the study, each individual child was assigned to a stratum according to that child's home address, thus allowing for a more accurate assignment of residential strata.
U.S. Census Bureau specifically defines urban and rural areas, however suburban
areas are defined by default, relative to the definitions of urban and rural
areas (Census of Population and Housing, 1990). Based on the U.S. Census Bureau
1990 definitions, "urban" is defined in terms of territory, population,
and housing units, and are considered to be places of 2,500 or more persons
living in incorporated or unincorporated areas included in urbanized areas.
An urbanized area comprises one or more places ('central places') and the adjacent
densely settled surrounding territory ('urban fringe') that together have a
minimum of 50,000 persons. The urban fringe generally consists of contiguous
territory having a density of at least 1,000 persons per square mile.
urban fringe also includes outlying territory of such density, connected to
the urban area or fringe, and either within 1.5 road miles of the urban core,
or within 5 road miles of the core but separated by water or other undevelopable
territory (Census of Population and Housing, 1990).
is defined by the U.S. Census Bureau (1990) as territory, population, and housing
units not classified as urban. Rural areas may be divided into "places
of less than 2,500" and "not in places," a category that is comprised
of rural areas outside incorporated and census designated places and the rural
portions of extended cities.
general rule to determine strata for this study was developed by the investigators
based on the two variables of population density and distance from the urban
center. Areas designated as being "urban" were within 2 miles of the
center business district. "Urban" also included areas that were between
2-3 miles of the center business district if the population density was 3,000
or more people per square mile. "Suburban" designation was assigned
to areas having a population density greater than 2,000 persons per square mile
and that did not qualify as being urban. "Rural" was considered to
be areas with a population density less than 2,000 persons per square mile.
of the influence of the Mississippi River on the geographic layout of Rock Island,
IL, the following definitions of residential strata for that population center
were based solely on population density: Urban was considered to be greater
than 3,000 people per square mile, suburban was between 2,000 and 3,000 people
per square mile, and rural was designated as areas having less than 2,000 people
per square mile.
of Elementary Schools
The method of sampling was a stratified cluster sample of school buildings located in the selected population centers. This sampling was accomplished by first contacting in writing the superintendent of each school district in the selected population centers. Along with a written explanation of the study was an invitation to participate during the course of this 2-year study. Receipt of a district superintendent's consent to participate served as permission to contact all of the principals of the school buildings in that district. A total of 41 districts were contacted; 21 (51.22%) superintendents consented to participate, 15 (36.59%) superintendents refused participation, and no response was elicited from 5 (12.19%) districts. It should be noted that only public school districts were sampled; there was no sampling of private schools or children being home schooled.
was described above, each participating school building was assigned a residential
stratum (urban, suburban, rural) based on their attendance zones. Once the individual
school buildings were sorted by population center and residential strata, buildings
within each stratum were assigned a number. Using a random number table, buildings
were selected to obtain a minimum total sample of 1,000 students in each of
the three strata across all population centers. For example, for the testing
conducted during Field Year 1, a minimum of 333 children were selected from
each rural, suburban and urban strata in each of the population centers of Des
Moines, Waterloo/Cedar Falls, and the Quad Cities. (Cedar Rapids provided us
with primarily an urban sample to supplement the urban sample from Waterloo/Cedar
Falls population center). This procedure was repeated for Field Year 2. Therefore,
because of the random sampling, some school buildings did not participate in
this study; some were selected to participate in only one year of the study;
and some school buildings were selected to participate in both years of the
field testing. Because the population of Iowa does not contain a substantial
number of African Americans, this sampling strategy was modified to over sample
the urban strata, since this stratum contained the largest proportion of African
1 presents the number of children who were sampled over the course of the study
as age-eligible participants according to the study site and strata.
1. Distribution of Participants by Strata and Study Site
EPIDEMIOLOGIC STUDY: TEST INSTRUMENTS
Phase Screening Instrument
The screening procedure only involved language performance. Children were not screened for hearing, nonverbal intelligence, or pervasive developmental disorder, the exclusionary criteria for SLI.
language screening test was developed that had a very high predictive relationship
with the diagnostic outcome. The screening tool consisted of 40 items from the
"Test of Language Development-2:Primary" (Newcomer & Hammill,
1988). This screening instrument was administered to each child individually
and took approximately 10 minutes to complete.
All data were entered directly onto computerized scan forms by the examiner during the screening. The University of Iowa Testing Services department then scanned these forms and transferred the data onto computer disk for analysis. See Chapter IV for the procedures used to assign the screening outcomes of pass or fail.
Phase: Diagnostic Battery
The goal of the diagnostic testing phase was to identify those children who would serve as SLI cases or control subjects. The diagnostic battery included hearing, language, speech, cognitive, and pre-reading tasks, and gross motor observations. Because the examiners who had conducted the screening also administered the diagnostic tests, the children had become familiar with the examiners. Thus, introductory warm-up sessions were unnecessary. Testing was administered in a standardized manner. All children participated individually, and diagnostic testing took approximately 2 hours to complete. The diagnostic battery was completely administered during one testing session; when this was not possible due to scheduling reasons, the testing was completed within a week of its initiation. However, individual tests, such as the TOLD-2:P and WPPSI, were always administered in their entirety during a single session. The order of administration of the diagnostic tests was held constant when possible. Within individual tests, the same presentation order of subtests was maintained across examiners. The examiners provided written comments regarding the testing situation and impressions of the child's performance to supplement the objective measures.
measures included in the diagnostic battery follow, along with a general discussion
of the testing procedures.
Because a hearing loss was an exclusionary criterion for the diagnosis of SLI, audiometric testing was performed. The purpose of the audiometric testing was to determine if the child had a persistent hearing loss that was suggestive of sensori-neural or conductive origins. Therefore, both pure tone audiometric screenings and acoustic omittance/impedance audiometry were conducted.
tone screening was conducted for 500, 1, 2, and 4 kHz at 20 dB (American Speech-Language-Hearing
Association, 1985). If the child failed the pure tone screening in an ear, pure
tone thresholds were obtained and a visual inspection of the ear canal was done.
Tympanometry was then done with four measures taken: Static Admittance (y A
passing range was .22 to .81); Ear Canal Volume (Vea passing range was .42 to
.97); Tympanometric Width (Gradient passing range was 59 to 151); and Tympanometric
Peak Pressure (TPP passing range was -139 to + 11). If anyone of the four measures
was failed in an ear, the child was considered to have failed the tympanometry
testing for that ear.
If a child failed the pure tone screening bilaterally, no further procedures were done at that time, and the child was retested, usually after a period of two weeks. If the child failed the pure tone screening unilaterally, the diagnostic testing was continued at that time.
the children who failed the pure tone testing unilaterally or bilaterally at
the first screening, a letter was sent to the parents/guardians to notify them
of the potential hearing problem and to suggest the appropriate audiologic or
medical follow-up as based on the results of tympanometry (see Follow-up Procedures).
If the child failed the second screening bilaterally, that child was disqualified
from further testing. Regardless of whether the child failed the second screening
unilaterally or bilaterally, a second follow-up letter was sent to the parents
to notify them of the audiometric testing results.
Because nonverbal cognitive ability was an exclusionary criterion for the diagnosis of SLl, nonverbal cognitive testing was performed. The Block Design and Picture Completion subtests of the "Wechsler Preschool and Primary Scale of Intelligence-Revised" (WPPSI; Wechsler, 1989) were administered. These subtests were chosen for two reasons: 1) of all the WPPSl performance subtests, these two were reported to be most highly correlated with full performance scale score (Block Design I = .59, Picture Completion I = .60), and 2) they afford an objective scoring method. These two performance subtests have been reported in the literature as a short form of the WPPSI Performance scale (LoBello, 1991). Further, these are easily administered and scored, and provide important criteria to assure inter-examiner reliability across the multiple field examiners.
scaled score for each subtest was reported. The two scaled scores were summed,
and a score of 16 or greater was selected for passing decisions. This summed
score reflects a performance intelligence score greater than 85.
Language testing was conducted to identify the SLI cases and control subjects. Subtests of the "Test of Language Development-2:P" (TOLD-2:P; Newcomer & Hammill, 1988) was supplemented with a narrative story task (Culatta, Page, & Ellis, 1983). These tests were selected because they assessed multiple aspects of comprehension and production, and provided normative data that allowed the calculation of standard scores.
these language measures were easily administered, and enabled development of
scoring guidelines for reliable scoring across the multiple field examiners.
The 5 TOLD-2:P subtests that were administered were Picture Vocabulary (PV),
Oral Vocabulary (OV), Grammatic Understanding (GU), Sentence Imitation (SI),
and Grammatic Completion (GC). The Word Articulation (W A) subtest was administered,
however these results did not contribute to the language diagnosis. The subtests
of the TOLD-2:P were administered according to the manual. Detailed scoring
guidelines were developed to assure consistency within and across examiners
during the field testing. Raw scores for each subtest were converted to standard
scores based upon local norms.
The Narrative Story task involved the retelling and comprehension of a short story about a birthday party. The examiner would read the story, and then ask the child to retell it. A maximum of three general recall were provided if the child needed prompting. The reported score was number of story events mentioned out of a possible 21.
number-of-events-retold score was supplemented by the examiner's rating of the
child's completeness and organization of recall. Completeness of recall was
rated on the basis of whether or not all major components of the story were
included in the retelling. To be considered complete, the child needed to include
(1) Setting and problem (it was the boy's birthday, he wanted a puppy, his mom
said "no"); (2) Complicating problem (the boy had a party and received
presents, but did not get a puppy); and (3) Resolution (the boy was surprised
and he got a puppy).
organization of recall was rated on the basis of whether or not the main components
of the story were retold in the correct sequential order, regardless of whether
or not it was complete. Examiners rated organization of recall with a "yes"
or "no." If the child retold no events or only one event, the organization
of recall rating was scored "not applicable."
the child's retelling of the story, the examiner asked the child 10 questions
to measure comprehension and memory of the story. The score reported was the
total number of questions correctly answered out of a possible 10. The raw scores
obtained for events mentioned during the retelling and for the comprehension
questions were standardized.
Reading Readiness was a variable of interest for the risk factor study. However, rather than probe the parents about Reading Readiness (such as asking about reading in the home and the child's exposure to reading), measure was obtained. The Letter Identification subtest of the "Woodcock Reading Mastery Tests-Revised" (Woodcock, 1987), the "Word-Sound Deletion Task" (Catts, 1991), and the "Random Animals-Colors Task" (Catts, 1991) were used to measure pre-reading skills.
The Letter Identification subtest is a measure of Reading Readiness. This test measures the child's ability to identify verbally letters written in several forms and scripts (such as upper or lowercase; roman, italic, and bold types; cursive and printed characters). Administration and scoring of this subtest was in accordance with the manual.
data for each month of the academic year are provided in the manual so that
scores obtained from children who are tested later in the year are adjusted
to control for learning that has occurred during the year. As specified by the
manual, raw scores were first converted into a "W" score. A difference
score was then calculated by subtracting from the W score a reference value
based on the month that the child was tested. The difference score was then
converted into a standard score, which is the reported score.
information for the Letter Identification subtest shows that the split-half
reliability coefficient for Grade 1 is of r= .94. The standard error of measure
for the W scores as reported in the manual for 1st grade scores is 4 W scale
The Word-Sound Deletion task (Catts, 1991) is a sound segmentation, phonological awareness task. In this task, the child was required to delete the initial phoneme or syllable from a word and repeat only the remaining phoneme or syllable. Three example items, all compound words, were demonstrated using pictures. For example, the directions were "Say 'baseball"' as pictures of "base" and "ball" were shown. Then, with the first picture covered, the child was asked "Now say 'baseball' without the 'base'." If the child did not respond correctly, the correct answer was provided for the demonstration items.
the child showed an understanding of the task, the testing began. The directions
were the same for each of the 21 test items, however, the pictured stimuli were
discontinued. The stimuli consisted of compound words, two syllable words, and
monosyllabic words. The sound sequence remaining as the correct response was
always a high frequency word. One repetition of an item was provided if needed,
and testing was discontinued when 6 consecutive items were incorrectly answered.
The reported score was the raw score, and the maximum raw score was 21.
The Random Animal-Colors task measured rapid naming ability, a skill which has been reported to be a measure of phonetic coding ability (Catts, 1991). The Random Animal-Colors task involved showing the child an 11.5" x 17.5" page that contained images of 24 animals. These 24 animals were 1 of 3 randomly selected animals (a pig, a horse, and a cow) that were colored in 1 of 3 randomly selected colors (blue, red, or black). These colored animals were arranged in random order in 4 rows of 6 items each. The child was first given as much practice as needed to identify the animals and colors, and several demonstration items were provided to allow the child to practice responding with "adj+noun" responses. If the child did not know his/her colors or animals, this task was not administered. The examiner instructed the child to "name these as fast as you can" in sequential order.
reported score for this task was the total time required for the child to name
all of the colored animals, measured with a stopwatch. Thus, a lower total time
score reflects better performance. A tally of incorrect responses was kept for
Word-Sound Deletion and Random Animals-Colors tasks were reported by Catts (1991)
to be the best combination of predictors of reading group membership for a group
of kindergartners that was comprised of 41 SLI and 30 control subjects. Catts
reported that these tasks together enabled them correctly to classify 82.9%
of the children in their study.
Severity Rating Scale
The Iowa Severity Rating Scale (ISRS; Jeffrey & Freilinger, 1986) was developed for use by speech-language pathologists working in the Iowa Public School system. The purpose of the ISRS measure in this study was to supplement the standardized measures of speech and language with a clinically significant measure, and to obtain an informal measure of voice and fluency.
ISRS is a 5-point severity rating scale of speech, language, voice and fluency
skills. The ratings are on a continuum, where 0 indicates adequate skills and
4 indicates a disorder. There are specified published criteria to guide the
rating made by the speech-language pathologist, and these guidelines were used
during this study (Jeffrey & Freilinger, 1986). Minimal additional guidelines
were established for use during the Field Study because the examiners' subjective
impressions were desired. Four guidelines were used:
Articulation: The Word Articulation subtest of the TOLD-2:P was supplemented
by the ISRS Articulation rating made by the examiner. A rating of "1"
indicated developmental s, r, l problems. Because certain phonemes (initial
k, m, v, n) are not sampled by the Word Articulation subtest, there was in some
cases a discrepancy between the WA results and articulation severity rating.
Language: The language severity rating was determined by the speech-language
pathologist based on observation of the child in informal interactions as well
as during the standardized language testing. In some cases, the TOLD-2:P raw
scores were converted to standard scores by the examiners, and this information
was considered when making the severity rating for language, as specified by
the ISRS manual. If the examiner did not have sufficient information to make
a language severity rating, a language sample was elicited. For children with
low performance scores on the WPPSI, the language severity rating was made commensurate
with cognitive ability.
Voice: The only guidelines provided to supplement the manual were that mild
hypernasality was rated as a "2," and moderate hypernasality was rated
as a "3."
Fluency: Ratings were made in accordance with the ISRS manual for fluency.
Two gross measures of motor skills were obtained: gait, and handedness. The examiner observed the child walking to the examining room, and noted if there were any obvious gross motor problems when walking. The child was also asked to write her/his name, and the examiner made note of which hand the child used.
Data Entry and Diagnosis
The diagnostic data were coded using a data entry program. To minimize coding errors, the data were entered using a two-pass method. Data were first entered and then verified during a second entry of these same data. Thus, data were entered at two different times and usually by two different people. The data entry program also had range checking for data having a minimum low and maximum high value.
outcomes were determined based on language performance as well as hearing and
nonverbal cognitive performance. Based on the diagnostic results, the children
were assigned to one of five diagnostic categories: (1) specific language impaired
(SLI; failed language testing but passed hearing and nonverbal cognitive testing);
(2) control (C; passed all language, hearing, and nonverbal cognitive testing);
(3) language impaired (LI; failed the language testing and nonverbal cognitive
testing, but passed hearing testing); (4) cognitive failure (CF; pass language
and hearing testing, but failed nonverbal cognitive testing); and (5) hearing
failure (HF; failed both hearing screenings and did not continue for further
Development of the Risk Factor Survey
A major research question in this study was concerned with identifying risk factors for SLI. In order to address this question, information was obtained from the parents of SLI children and non-SLI controls regarding a wide range of exposures to potential risk factors. This information was obtained through a telephone survey that lasted approximately 50 minutes.
candidate risk factors were selected after a literature review of risk factors
and language impairment. A pilot survey was developed to obtain information
regarding these risk factors, and was administered to 6 parents. Several modifications
were made based on the pilot. A data coding manual for the final Risk Factor
Survey is contained in Appendix I. The general categories of risk factors selected
for this study are:
Factor Survey Administration
The names of the parents who were to receive the Risk Factor Survey were sent from the lab at the University of Iowa to the Statistical Laboratory at Iowa State University (ISU), a subcontractor on this contract. The lab at the University of Iowa contacted these parents by letter to notify them that they had been selected to receive the Risk Factor Survey. The lab sent a list of chemicals for the parents' reference during the telephone call. Trained professional interviewers at the ISU Statistical Laboratory then called the parents to arrange a mutually convenient time to administer the survey. The interviewer usually administered the survey during one call. The parents almost always received the telephone survey prior to receiving any information regarding the outcome of their child.
supplementary information booklet was developed to aid the telephone interviewers
and to assure standardized administration of the survey. This booklet explained
terms used in the survey, as well as general instructions for the interviewer,
on a question-by-question basis. All interviewers were blind regarding the diagnostic
outcome of the study child.