Goldman-Fristoe Test of Articulation 3

Can you use either the cartoon pics or the realistic pics with any age and can still use the standard scores?

Because the test stimuli are the same for all ages of examinees, you may use either the cartoon pictures or the realistic pictures with a child of any age and still use the standard scores. The set for children ages 2:0 through age 6:11 were selected for that age because the art is designed for younger children. Children as young as age 7:0 showed a preference for the more realistic art. 

What happened to blends /fl/ and /kl/?

During the development process, the authors decided that not all blends needed to be included in the test. In addition, other singleton consonants such as the “zh" sound, as in "measure" is not tested.

Why is the word "yellow" tested when none of my clients can produce this word correctly--they always say "lello" or "yeyo".

The word "yellow" is included for several reasons:

1. it is word that is consistently and easily elicited from children,
2. it is one of two words that is a more complex context for /l/--along with "elephant,
3. it is a common word that is correctly produced by most typically-developing children by age 4 1/2 to 5, and
4. it is a word that has differentiation between typically developing children from those who have a disorder.  Younger children would not be penalized for saying "yeyo" or "lelo" because it is a common substitution for them, especially if they produce /l/ correctly in the simpler contexts on the test.

Why isn't the medial voiceless /th/ and the medial "j" as in orange tested?

Medial "j" is tested in Items # 39 and 50 in Sounds-in-Words (pajamas, vegetable); medial voiceless "th" is not tested. In the Sounds-in-Sentences Story 1, medial "th" is tested in "something". Some words are difficult to represent in a spontaneous naming task.

On GFTA-3, is the vowelized r (like in the word "card") addressed separately from vowelized (like the word "Mary")

On GFTA-3, all /r/ productions are scored as correct or incorrect. You are able to view the contexts in which an examinee can correctly or incorrectly produce consonantal and vocalic /r/ using the Vowel Error Analysis in the Record Form.

Is it OK to use the younger age story for Sounds-in-Sentences with lower-cognitive-ability students?

If you want to obtain qualitative information about an examinee's skills, you may administer the Sounds-in-Sentences test for ages 4:0 through 6:11 to students ages 7 and above. On page 11 of the Manual, you will find a brief summary about Out-of-Age-Range testing administration.   Out-of-age-range scores are not available for GFTA-3 (GFTA-3 Manual, page 22). The Sounds-in-Sentences stories have a different number of test items and different content, and the tests were not normed with individuals outside of the target age range.

Why are Sounds-in-Sentences scores sometimes higher than the Sounds-in-Words scores?

The Sounds-in-Words and Sounds-in-Sentences tests provide information about an individual's accuracy of production, but in different contexts. The Sounds-in-Words test provides information about spontaneous productions of single words; Sounds-in-Sentences provides information about productions of imitated sentences. The differences in the two scores indicate the difference in requirements of the task and the performance differences of the normative sample when completing the two tasks.

Requirements of the task:

Sounds-in-Words: spontaneous production of single words.

Sounds-in-Sentences: imitated productions in sentences.

Performance Differences in the Two Tasks.

Even though the word task is based on spontaneous productions and the sentence task is based on imitated productions, the word task is an “easier” task than the sentence task—typically developing children make proportionally fewer errors on Sounds-in-Words than on Sounds-in-Sentences.

The raw scores are not comparable. Children have fewer opportunities to produce individual sounds on the Sounds-in-Sentences task so the total error scores are very different from the Sounds-in-Words task.

Although the standard scores for Sounds-in-Words and Sounds-in-Sentences are on the same scale (mean of 100, SD of 15), the differences in the performance of the normative sample on the two tasks are apparent in the standard scores for the 7-year-old girl tested.

Sounds-in-Words  A female who is 7:11 who cannot produce a single “-er”, /r/, or /r/ blend in any single word context performs extremely low compared to peers. A standard score of 45 indicates that almost all females at age 7 were able to produce /r/ in almost all contexts—to miss every single /r/ at that age is grossly atypical.

If you refer to the Manual, Table D.1 on page 238, you will see that 50% of children age 2:0 through 2:5 in the GFTA-3 normative sample produced "er" one or more times correctly. About 75% of children 2:0 through 2:5 produced /r/ correctly one or more times, and 90% of children ages 2;6 through 2:11 produced /r/  one or more times correctly. These ages represent emergence of correct productions, not mastery. Even so, these numbers show that even very young children can produce an /r/ or "-er" correctly in one or more contexts. From age 2 through age 6, an increasing percentage of female children produce /r/ correctly.  Mastery of the sound (90% of the normative sample producing a phoneme correctly with at least 85% accuracy) occurs later. Table D.2 shows that females master /r/ and /r/ blends from age 6:0 to 6:11. Because the child you tested did not correctly produce a single /r/, her performance is significantly different from same age, same sex peers.

Sounds-in-Sentences  In contrast,  a female who cannot produce a single “-er”, /r/, or /r/ blend in Sounds-in-Sentences still performs low (SS of 79, just short of 1 ½ SDs below the mean), but not extremely low compared to other 7 year old girls.

Why are the standard scores so much lower on the Sounds-in-Words and Sounds-in-Sentences tests in a case where a 7-year-old girl misses every /r/ on both tasks?

Even though Sounds-in-Sentences is an imitation task, producing sentences is expected to be more difficult than producing single words.

The standard scores show that in the sentence task, even when a child misses every single /r/, she is not performing as poorly on the Sounds-in-Sentences task as she did on the Sounds-in-Words task compared to same age, same sex peers. This means that while most 7-year-old girls produce /r/ sounds correctly in sentences, there are some girls who make a few errors in sentence productions. The 7-year-old girl’s performance (while still poor) did not result in a score as low as in Sounds-in-Words, a test in which girls much younger than the child you tested would be expected to produce at least a few  /r/ correctly in some contexts.

Is there any chance that the test protocol will be altered to make it more user friendly is sharing data and results with the parents?

An updated Record Form became available for all new kit and record form purchases beginning in late July 2016. The new record forms were included in all kit and record form package orders, along with instructions and additional information about interpreting test results on GFTA3.com (see the Resources tab).

Do "r" and "s" distortions count as errors?

A distortion of any phoneme is counted as an error. If the sound the child produces is acoustically accurate, but the articulator placement is not quite on target, the phoneme is not counted as an error.

How do you score a production in which the child presents with a forward tongue protrusion/tongue thrust?

If the sounds produced are distorted, these are counted as errors. If the tongue placement error does not negatively impact perception of the sound (it is acoustically accurate), note the tongue protrusion in your report, but do not count the incorrect placement of the tongue as an error.

How do you suggest scoring for distortion errors of sibilants, such as lateralized /s, z/ (lateral lisp) or interdentalized or dentalized /s, z/ (frontal lisp)...or lateral distortion of "sh", "ch", and "j"?

All distortions are counted as errors (see response above).

How do you take into account if a child presents with a forward tongue protrusion / tongue thrust? 

If the sounds produced are distorted, these are counted as errors. If the tongue placement error does not negatively impact perception of the sound (it is acoustically accurate), note the tongue protrusion in your report, but do not count the incorrect placement of the tongue as an error.

If a child has a significant number of dentalized errors (due to tongue thrusting) on the GFTA-3, do these errors count toward the raw score or are they only noted on the form, but not counted as a raw score error? In other words, are the dentalized errors considered a diacritical mark and therefore, not counted in the raw score or are the dentalized sounds considered a distortion and therefore counted in the raw score?

If the sound production is distorted by dentalization or there is a substitution error such as th/s, you would score the response as an error and count it toward the child’s raw score. If the child dentalizes a phoneme, but the production “sounds OK, but placement is off slightly”, you might choose to note the child’s placement with a diacritical mark, but not count the production as a speech sound error.

Why does GFTA-3 penalize preschoolers by giving them  so many /s, l, r/ occurrences. This is problematic.  They are not penalized as heavily for /k, g, f/ errors which we would expect in many of our preschoolers. 

The GFTA-3 normative scores take into account that preschoolers have not mastered later developing sounds such as /r, s, l/. Beginning at age 2, the percentage of typically developing children who are producing these sounds in a one or more contexts ranges from 50% (for -er) to 90% (for /s/ and /l).   The percent of correct production of these sounds increases with age. This developmental pattern is seen long before the age of mastery (where 90% of children produce the sound correctly at least 85% of the time on the GFTA-3). This developmental pattern is built into the normative scores. Very young children are not expected to produce 85% of /s, l, r/ sounds correctly, but the data show that there should an increasingly high percent of correct productions long before mastery at ages 7 and 8. 

Are there different norms tables for examinees who speak a variation of English? And if so, are there tables describing dialectical differences in the manual?

Appendix E in the Examiner's Manual provides common phonemic contrasts for speakers of African-American English and Spanish-influenced English. Common phonological patterns that may be seen for speakers of Asian-influenced English are also included. When the child is identified by the clinician (through parent interview, evaluation of dialect, observation of the child's speech patterns) as being a speaker of a dialect other than Standard American English, responses that reflect a dialect variation are scored as correct. If you listen to the May 11, 2016 GFTA-3 webinar posted online, the presenter's explanation of Slide 20 is correct, but the slide text is incorrect--the incorrect text reads "Dialectal variations are counted as errors in GFTA-3 norms." The text should have read "Dialectal variations are NOT counted as errors in GFTA-3 norms." When scoring GFTA-3, do not count sound productions that are dialectal variations as errors. The slide text error will be corrected in the 9-15-16 presentation "Identifying Typical and Atypical Speech Patterns Using the GFTA-3" available online in mid-September.

Are dialectal variations counted as errors? That is what it says on the slides posted online.

That is an error on the slide. Dialectal variations are NOT counted as errors on GFTA-3 (see Manual, pages 10-11). The voice recording online is correct, but the slide text is not. The slide text error will be corrected in the 9-15-16 presentation "Identifying Typical and Atypical Speech Patterns Using the GFTA-3" available online in mid-September.

How should we account for sound additions? If a child says "drump" for "drum," do we count the final /m/ correct or incorrect? Q-global does not allow us to account for sound additions.

You can mark additions for qualitative information, but GFTA-3 does not account additions as errors. You mark correct production of the target sounds. Additions of sounds sometimes show up in speakers of English influenced by another language.

How can a child's score on GFTA-2 be so different on GFTA-3?

The developmental ages for speech sound mastery has not changed significantly over time. The GFTA-3 data showing the ages at which children demonstrate one or more correct productions of a phoneme are comparable to the ages listed in the GFTA-2 Supplemental Developmental Norms booklet. However, on GFTA-3, children with certain error patterns and at certain ages may earn scores much lower than their same age/gender peers on they would have on GFTA-2.

So what is causing the differences in scores?

GFTA-2 Premise: if a phoneme is counted as correct, the child has mastered production of that phoneme.

The GFTA-2 assesses one occurrence of a sound in each position of a word (if applicable). If a child makes an error on a sound not targeted in that word, the error is not counted toward the raw score. A GFTA-2 raw score does not include errors in other phonetic contexts. Performance on GFTA-2 reflects emergence of phoneme production (one instance) rather than mastery (correct production in 85% or more contexts).  

If you used GFTA-2 and have the Supplemental Developmental Norms book, you can see on page 8, Sound number 17 that 30% of children 2:0 to 2:5 produced /r/ correctly in initial position, 57% of children in medial position, and 57% in final position. Those percentages increase across age groups until you get to about 85% of the children correctly producing that /r/ at age 5 1/2 and 95% of children at age 7 producing /r/ in all positions.

GFTA-3 Premise: Phonemes should be tested in multiple contexts because productions can be affected by surrounding vowels and consonants and the complexity of the word structure. In addition, research shows that there is a range of phoneme acquisition, with a time period between emergence and mastery of sounds. Phoneme mastery occurs gradually, with increasing ability to produce the sound in multiple and more complex contexts.

On GFTA-3, all errors are counted toward the raw score. Inclusion of multiple occurrences of /r/, for example, does not penalize a young child who is not expected to master /r/. Data collected during standardization show that /r/ is produced correctly by 90% of children in one or more contexts (not mastered) starting at age 2:6 to 2:11 (males) and 3:6 to 3:11 (females).  Young children who demonstrate a few instances of a correct /r/ production will earn standard scores in the average range.

Low standard scores indicate that a child’s speech is not comparable to age/gender peers. Table D1 shows the ages at which 50%, 75%, and 90% of typically developing children use each phoneme in one or more contexts. You can compare the information to the examinee’s performance. Table D2 shows age of mastery for each sound by gender. For example, males master /r/ in initial and final position of words at age 7:0 to 7:11; females master /r/ at age 6:0 to 6:11. If an older 6-year-old child does not produce a single /r/ correctly on GFTA-3, this is atypical compared to same age/gender peers, and the child will earn a very low standard score.  A norm referenced test like GFTA-3 gives you information about how a child performs compared to peers. Your other assessment data (stimulability, performance of children in your area, and your clinical expertise) guide your decision to re-evaluate the child later, monitor the child, or enroll the child in therapy.

Please explain the criterion applied to indicate "emergence" vs. "mastery" of speech sounds.

On the GFTA-3, "emergence" of speech sounds is defined as the age at which 50%, 75%, and 90% of the children in the GFTA-3 nromative sample produced phonemes corectly one or more times during test administration.  "Mastery" of speech sounds indicates the age at which 90% of the children in the GFTA-3 normative sample articulated the phonemes in initial, medial, and final positions of words with at least 85% accuracy.

The developmental chart I use shows the ages of speech sound mastery as different from both Tables D1 and D2. Why is that?

Developmental charts have been developed based on research conducted across different years (from the 1960s through the present), with somewhat different mastery criteria (e.g., 85%, 90%, or 95%), and with widely disparate populations (e.g., primarily mainstream populations, excluding individuals who speak variations of Mainstream American English or individuals who are English learners to diverse populations that did not exclude these individuals.)  The beginning of the age band for phoneme acquisition on a developmental chart that shows a range of ages for speech sound acquisition is comparable with GFTA-3 Table D1--it shows the age at which a certain percentage of children begin to produce a phoneme correctly. On GFTA-3, you can see the percentage of children who produce a phoneme correctly for a given age. The end of the age band on a developmental chart that shows a range of ages for speech sound acquisition is comparable to GFTA-3 Table D2--it shows that age which almost all children have mastered a phoneme.

On Table D.1, is it saying that 90% of students master R in the 2-6 to 2-11 age range or just beginning to make R sounds?

Table D1 indicates the age at which 50%, 75%, and 90% of children in the GFTA-3 normative sample produced phonemes correctly one or more times during test administration. It does not indicate mastery of the phoneme. The age at which mastery is demonstrated (defined on GFTA-3 as the ages at which 90% of the children in the GFTA-3 normative sample articulated the phonemes in initial, medial, and final word positions with at least 85% accuracy.

Were there any bilingual speakers included in the norming sample (i.e., non-monolingual English speakers)? Were these children exposed to English from birth or was there a variety of first language exposure?

In the GFTA-3 normative sample, 13% of the normative sample were bilingual. To be included in the normative sample, individuals who identified as being bilingual were required to be simultaneous language learners (learned English and the other language simultaneously), and reported that both languages were learned in the United States. In addition, English had to be the child's most frequently used language.

Will there be a KPLA-3 to accompany the GFTA-3?

KLPA-3 is currently available (published in 2015).

Could you give a brief overview of the differences in the types of digital/electronic scoring?

If you use Q interactive, you obtain scores automatically with the press of a button once testing is concluded. Q-interactive is an interactive assessment that uses two iPads connected via Bluetooth. You use one iPad to access test administration directions, score and record responses, and control visual stimuli. At the end of the test, you automatically obtain test scores and a score report. To use Q-interactive, you would purchase a yearly license that gives you access to CELF-5, GFTA-3, KLPA-3, and PPVT-4. A usage fee is charged each time you administer a test to a client--this fee includes administration of the test to the child, access to the Manual, and automated scoring.  If you use Q-global, when you complete testing, you would use your purchased paper Record Form and enter either raw score totals or individual item scores in the Q-global platform to obtain a score report. You may purchase individual score reports or a 1-, 3-, or 5-year subscription. See helloQ.com for more information.

How should we account for sound additions on Q-global? If a child says "drump" for "drum," do we count the final /m/ correct or incorrect?

If the child produced /m/ correctly, you would score final /m/ as correct. You may choose to mark additions of phonemes on the Record Form for qualitative information (noting the addition of /p/), GFTA-3 scoring rules do not account for additional phoneme productions, and the addition is not counted as an error.

Q-global does not have a place to notate additions of sounds.

Because the raw score is calculated based on the number of misarticulation of target sounds, phoneme additions are not represented in the raw score. You can mark additions in the Notes section for qualitative information, but GFTA-3 does not account additions as errors.  Additions of sounds sometimes show up in speakers of English influenced by another language.

How long of a time period can a clinician use GFTA-2 because switching to GFTA-3?

Recognizing that there are many variables that can influence a decision to update to the newest edition of a test (e.g., test content and design, funding, appropriateness/acceptance of test results based on outdated editions of tests), the Standards for Educational and Psychological Testing (2014) does not specify a required time interval.  While that is the case, local government, employers, agencies, or funding entities often establish requirements for using the most current editions of tests. Best practice guidelines suggest that clinicians should use the most recent test edition available because the most recent version incorporates current best practices in assessment, current research, and norms based on a sample that matches the most recent demographic characteristics of the population. It is now common that tests in a clinician's toolbox are updated within 1-2 years following publication of the new edition.