Free tool · Phonology
Percent consonants correct (PCC)
Calculate PCC from a transcribed speech sample. Start with dictionary-generated IPA targets, edit them to reflect the child’s productions, and check each utterance against your recording. Review the score and consonant-by-consonant results below.
- Paste your transcriptEnter the child’s utterances to look up IPA targets.
- Check each wordReview targets and productions against your recording.
- Review errorsSee substitutions, omissions, and additions for each word.
- Review PCCSee PCC and consonant counts for the utterances you checked.
- Save your worksheetDownload your work to keep a record or resume later.
No account or upload is needed. Download your worksheet in step 5 before closing or reloading this page; your work is not saved automatically.
1 Paste the child’s utterances
One utterance per line. A speaker label at the start of a line (C:, CHI:, *CHI:) is ignored. To resume a saved worksheet, paste its contents here.
Preparing the transcript for PCC
Prepare your transcript using these sampling rules from Shriberg and Kwiatkowski (1982, Appendix B):
- Write
xxxfor a word you cannot make out, or can only partly make out, or whose gloss is a guess. It is left out of the score and counted toward intelligibility. - Write a stuttered word once: ba-balloon is balloon. Only the first attempt at a syllable is scored.
- Keep at most two of a word repeated in a row (cat cat cat is cat cat), unless the child said it differently each time.
- Write the word the child meant. A dialect form, a casual reduction (dono for don’t know) or an allophone (the flap in water) is correct under these rules, so the target in step 2 should be the child’s target; change it there when the dictionary’s form is not.
- Leave out fillers (um, uh).
2 Check each word
Review each IPA target and adjust it as needed. The production box initially matches the target; edit it to reflect what the child said. Fill in missing targets for words outside the dictionary. After checking an utterance against your recording, mark it checked to include it in the score.
3 Errors
Words with differences between target and production appear here for checked utterances: ɹ→w replaced, k→∅ omitted, ɹ+k addition to a consonant. Grey consonants are scored as correct.
4 PCC results
Results include only utterances marked checked in step 2. Review the totals and scoring rules below.
Scoring rules
PCC is consonants correct divided by consonants attempted, as a percentage, scored by the rules in Shriberg and Kwiatkowski (1982, Appendix B). The scoring details are described below.
The target
Each word’s target is its citation form in General American English from the CMU Pronouncing Dictionary: the word said on its own, one pronunciation per word, no stress or length marks. Adjust the target for the child’s dialect and speech context (see Preparing the transcript in step 1) before scoring.
Which sounds count
The 24 English consonants: p b t d k ɡ m n ŋ f v θ ð s z ʃ ʒ h l ɹ w j, with tʃ and dʒ as one consonant each. Post-vocalic ɹ as in fair and car is a consonant; the rhotic vowels ɚ and ɝ as in water and bird are vowels, and vowels are not scored. Shriberg et al. (1997) added a separate vowel metric (PVC) for them.
Correct or incorrect
A consonant is correct when the produced form has it, in order, written the same way. Anything else is incorrect: a deletion, a substitution (a glottal stop or a cognate included), a distortion, which Shriberg counts “no matter how subtle,” or an added sound on the consonant (cars as kɑɹks makes the ɹ incorrect). An added sound beside vowels only (on as hɑn) is not scored, since there is no target consonant. PCC-R (Shriberg et al., 1997) scores distortions as correct; this page reports PCC. The produced consonants are lined up against the target’s in order, so pun for spoon scores the p and n correct and the s deleted. ∅ in place of a consonant marks it as deleted.
Context-dependent scoring: Shriberg counts a dropped initial h or a final ŋ said as n as an error only in a stressed syllable, and partial voicing only at the start of a word. Apply these exceptions manually: enter the target form in the production box when the rule treats the production as correct.
Severity ranges
Shriberg and Kwiatkowski (1982) set four ranges from the conversational speech of 30 children with delayed speech, aged 4;1 to 8;6 (mean 5;9), rated for severity by a panel of clinicians:
| 85 – 100 | mild |
| 65 – 85 | mild to moderate |
| 50 – 65 | moderate to severe |
| below 50 | severe |
They apply to a conversational sample scored by the rules above, from a child like those; Shriberg, Austin, Lewis, McSweeny and Wilson (1997) say not to convert a PCC from any other kind of sample into them. A short sample, a different task, or a much younger or older child is a different measurement.
5 Save your worksheet
Copy or download your worksheet to save the targets, productions, and review status for each word. To continue later, paste the saved text into step 1.
The calculator is just the beginning.
Arc is being built to keep pronunciation review alongside your recording and transcript—and carry the results into your report.
Join the alpha
Sources
- Shriberg, L. D., & Kwiatkowski, J. (1982). Phonological disorders III: A procedure for assessing severity of involvement. Journal of Speech and Hearing Disorders, 47(3), 256–270. https://doi.org/10.1044/jshd.4703.256
- Shriberg, L. D., Austin, D., Lewis, B. A., McSweeny, J. L., & Wilson, D. L. (1997). The Percentage of Consonants Correct (PCC) metric: Extensions and reliability data. Journal of Speech, Language, and Hearing Research, 40(4), 708–722. https://doi.org/10.1044/jslhr.4004.708
- Carnegie Mellon University. The CMU Pronouncing Dictionary (cmudict). https://github.com/cmusphinx/cmudict. Used under its BSD-style license; see the notice.