Note: For a complete list of substitutions (the target words in GFTA-2 used to sample sounds that were in dropped GFTA words), go to GFTA-2 Conversion Table for Use with KLPA.

You really cannot compare the two scores because the GFTA and the GFTA-2 are two very different tests.

The content is different:

• The total number of sounds evaluated in Sounds-in-Words is different. There were 73 items (sounds sample) for the original GFTA. For GFTA-2, there are 77.
• Different sounds are evaluated in each test. The GFTA-2 evaluates six consonant clusters (blends) that were not included in the original. The "hw" and the "skw" sounds were dropped in GFTA-2 because of infrequent use in the common core vocabulary of young children.
• There also seemed to be a question among speech-language pathologists whether to count the blends in the original GFTA. Many did not. The blends are definitely counted in the GFTA-2. This could account for a discrepancy between a score "on file" and a GFTA-2 score.

The sampling and normative procedures are different:

• The two tests were standardized on very different populations. The norms for the GFTA-2 are based on children currently living in the United States. The GFTA norms are over 30 years old.
• The age range on GFTA-2 for normative interpretation was expanded to 2-0 through 21-11 to meet the requirements of Public Law 94-142 and Public Law 99-457 (now incorporated into the Individuals with Disabilities Education Act [IDEA] reauthorized as Public Law 105-17).
• To be in full compliance with IDEA, the GFTA-2 sample includes special education students (approximately 10 percent of the total sample) in the same disability proportions as reported to the U.S. Department of Education. This ensures that the full range of abilities were sampled and tested. Obviously, developing an articulation test on all "normal" children would result in both inappropriate and misleading normative data. Because the original GFTA norms were completed prior to P.L. 94-142, it is unclear whether children with disabilities and impairments were included.
• Because of advances in statistical methods and computer capabilities, we were able to sample separately by gender (male and female) and simultaneously control for race/ethnicity, region, and SES (mother's education level) during the GFTA-2 standardization. This means that the GFTA-2 sample closely matches the U.S. population according to recent census data.
• Percentiles and age-equivalents are not comparable from test to test because they are both based on an ordinal scale of measurement.
• Percentiles are a ranking of all scores along a continuum (first, second, third, etc.) They are not based on an equal-interval scale (that is, there is not an equal distance between first and second, second and third, third and fourth, etc.). Therefore, where the 20th percentile is on one test could be a very different place from where the 20th percentile is on another.
• Age-equivalents are based on the median raw score, which is the 50th percentile. Therefore, they are also an ordinal scale and cannot be compared test to test.

An age-based standard score is a "standard" way of representing an examinee's raw score performance that takes into account two things:

1. How far the individual's performance was from an average performance for his or her age. Was it above average, below average? How far above? How far below?
2. How much did children the same age vary in performance? How unusual was a low score for that age? How unusual was a very high score?

There are two basic ways of developing age-based standard scores. The first method maintains or preserves the shape of the distribution of raw scores for each age. The second converts the shape into a normal distribution (the bell curve).

Intelligence and vocabulary are two abilities that are normally distributed in the general population. Raw scores from these tests would form a normal distribution-look like the bell curve. In other words, few children score high, few score low, and most are bunched in the middle around the average raw score. Age-based standard scores for these tests would be developed using the second method.

Articulation ability is not normally distributed in the general population at most ages. After age two or three, GFTA-2 raw scores form skewed distributions-most scores are similar and bunch together with a long tail of infrequent scores. For example, most children over the age of 6 make only 0, 1, or 2 errors with fewer children making a lot of errors. Therefore, age-based standard scores for an articulation test must be developed by the first method-the one that preserves the shape of the raw-score distributions.

Using either method, the resulting age-based standard scores indicate how different the child is from average for his or her age and how unusual or rare the particular performance is.

For further information on interpreting GFTA-2 scores, go to ags.pearsonassessments.com/assessments/gfta_scores.asp.

Remember that the age-based standard score indicates the "distance" from average or how different the child is from average for his or her age. This may be the best way to demonstrate his or her need for services.

You are thinking of a normalized distribution. Usually on an intelligence or achievement test, students with extremely high scores fall in the range of three to four standard deviations above the mean (about 1 to 3 percent of the population). These scores would be at the top of the percentile, meaning that the student with a percentile of 99 has performed as high as or higher than 99 percent of the normative sample at that age. Another way to say it is that 1 percent or less of the sample performed better than the student did.

Because articulation ability is not normally distributed, percentiles cannot be interpreted in quite the same way. By the age of 8, many children have mastered articulation. A female aged 8-3 who has made no articulation errors has a percentile of >48. This means that she has performed as high as or higher than 48 percent of the normative sample at that age. In other words, more than half of the girls her age in the sample made no errors. When you look at a male who is 13-0 with no errors, his percentile rank is >24. More than 75 percent of the boys his age made no errors.

You need a copy of the GFTA-2 Supplemental Developmental Norms booklet. This booklet includes an extensive data set for order of consonant and consonant cluster development, based on the GFTA-2 nationwide standardization sample. The data set is presented in three different tables, with the standardization information addressed totally and then separately for males and females. The tables contain p-values for each initial, medial, and final consonant or consonant cluster sound addressed in the GFTA-2 Sounds-in-Words (77 in total). A p-value is the percentage of correct responses to an item or task, in this case, the percentage of correct articulations of a particular consonant sound.

The order of the data matches the GFTA-2 Response Form recording grid for Sounds-in-Words. The “Sound Number” in the first column of each table matches the row number of this grid. For example, Sound Number 4 is for /w/, which is only tested in the initial position. Reading across Table 1 for Total Sample, 79% of children aged 2-0 to 2-5 correctly produced the sound at the beginning of the target word “window.”

Using these tables for males and females, clinicians can set their own mastery age as defined as the youngest age at which the p-value is at a certain point (for example, .90 or greater) for each consonant or consonant cluster sound. Clinicians can then establish their own developmental criteria for each individual case or for their particular clinical setting based on this controlled GFTA-2 standardization sample. This empowers organizations to set their own cutoff criteria as to who qualifies for intervention services, based on developmental data.

By the way, we are supplying the GFTA-2 Developmental Norms free of charge to practitioners since this is such an important advancement to the field. [The Supplemental Developmental Norms were a research presentation at the 2000 ASHA Convention in Washington, D.C.]. Just call us at 800-627-7271 and ask for Customer Service.

Request your free GFTA-2 Developmental Norms online. Enter FREE NORMS BOOK in the comment area of the online form.