The Goldman-Fristoe Test of Articulation (GFTA-3) is scored by recording each speech sound error a person makes during the Sounds-in-Words and Sounds-in-Sentences subtests, classifying those errors by type, and then tallying them into a total raw score that converts to standard scores using age-based norms. The process is straightforward once you understand the error notation system and a few special rules for clusters, vocalizations, and dialectal variations.
The Three Subtests and What Each Measures
The GFTA-3 has three components, each scored slightly differently. Sounds-in-Words is the core subtest: the individual names pictures, and you record errors on specific target sounds in the initial, medial, and final positions of each word. Sounds-in-Sentences uses two short stories with target words embedded in connected speech, letting you see how articulation holds up in more natural language. The Stimulability subtest is administered last and only for sounds that were misarticulated during the first two subtests. For stimulability, you model the misarticulated sound in a syllable, word, and sentence, and the individual imitates your productions.
How to Classify Errors: Substitutions, Omissions, and Distortions
Every error you record on the GFTA-3 falls into one of three categories that count toward the total raw score, plus one category that does not.
- Substitutions: The person replaces a target sound with a different sound. Mark through the misarticulated sound in the IPA Transcription column and write the substituted sound in the appropriate position column (Initial, Medial, or Final).
- Omissions: The person leaves out a target sound entirely. Mark through the omitted sound and write a dash (—) in the position column.
- Distortions: The person produces something recognizable as the target sound, but it’s acoustically off. These count as errors. Write the sound along with a diacritic mark in the position column, or add a written note like “lateral release.” One important distinction: if the articulators are slightly misplaced (such as dentalization) but the sound produced is still acoustically accurate, that is not counted as an error.
- Additions: The person tacks on extra sounds that aren’t part of the target word (for example, saying “swings” instead of “swing”). Note the response, but do not mark anything in the position columns. Additions are not counted as errors and are not included in the total raw score.
Recording Errors on the Record Form
For both Sounds-in-Words and Sounds-in-Sentences, the recording process follows the same basic steps. When you hear an error, mark through the misarticulated or omitted sound in the IPA Transcription column on the record form. Then move to the adjacent Initial, Medial, or Final column and mark through the target sound there, writing in what the person actually produced (a substituted sound, a dash for an omission, or a diacritic-marked sound for a distortion).
If a response contains multiple errors, transcribe the person’s full response in the Response column. This gives you a complete record to refer back to and makes your error analysis more reliable.
Special Scoring Rules
Consonant Clusters
Each consonant cluster on the GFTA-3 is scored as a single unit. Whether the person drops one sound in the cluster or misarticulates the entire cluster, you count it as one error, not two or three.
Final /l/ and /r/ Vocalizations
When someone vocalizes a final /l/ or /r/, producing only a vowel sound instead of the expected vowel-plus-consonant combination, you score it in a specific way: mark the vowel preceding the /l/ or /r/ as omitted and mark the /l/ or /r/ as substituted by the vowel. This captures both the lost consonant and the changed vowel quality.
Adjacent Medial Consonants
When two consonant sounds sit next to each other in the middle of a word but are separated by a syllable boundary, score them as separate sounds rather than as a cluster.
Self-Corrections
If the person corrects themselves before you move on to the next item, write “SC” next to the originally recorded error. Transcribe the corrected response in the Response column so the final production is clear.
No Response
If the person doesn’t respond at all, mark through all target sounds in the IPA Transcription column and write “NR” in the Response column. Then mark through each sound in the position columns and write a dash for each one. Every target sound in that word counts as an omission.
Dialectal Variations Are Not Errors
This is a critical scoring rule. Accepted dialectal or regional variations of Standard American English are scored as correct responses. If you recognize a production as a dialectal variation rather than a clinical error, transcribe the response in the Response column but do not mark anything in the Initial, Medial, or Final columns. The sound is not counted toward the raw score. The GFTA-3 manual includes a full appendix (Appendix E) with examples of accepted dialectal variations you can reference during scoring.
Calculating the Raw Score
After you’ve finished recording all errors, count the total number of marked errors across the Initial, Medial, and Final columns. Each marked position column entry counts as one error. Remember that additions do not count, consonant clusters count as one error regardless of how many sounds are affected, and dialectal variations are excluded. This total gives you the raw score for each subtest.
Completing the Phonetic Error Analysis
Beyond the raw score, the GFTA-3 includes a Phonetic Error Analysis that organizes errors by individual sounds. This analysis feeds into two supplemental tables. Table D.1 tracks phoneme emergence: you circle any phoneme the person got wrong in every occurrence across the test. Table D.2 tracks phoneme mastery using a stricter standard: you circle a phoneme if even one instance was produced incorrectly in a given word position. For both tables, you then draw an “age line” under the person’s age range to see which errors fall within developmental expectations and which fall below. If the person is 9 or older, draw the line under the oldest age range listed.
These tables give you a quick visual comparison between the person’s error patterns and typical development. Sounds circled above the age line are ones you’d generally expect to still be developing. Sounds circled below it suggest the person is behind where most peers their age would be.
Converting Raw Scores to Standard Scores
The raw score from Sounds-in-Words is the primary score used for norm-referenced interpretation. You convert it to a standard score using the age-based tables in the GFTA-3 manual. The standard score tells you how the person’s articulation compares to others of the same age. The Sounds-in-Sentences subtest provides a supplemental comparison of articulation in connected speech, while the Stimulability results help guide treatment planning by showing which misarticulated sounds the person can already produce with a model.